Scaling and fouling in membrane distillation for
desalination applications: A review
The MIT Faculty has made this article openly available. Please share
how this access benefits you. Your story matters.
Citation
Warsinger, David M., Jaichander Swaminathan, Elena GuillenBurrieza, Hassan A. Arafat, and John H. Lienhard V. “Scaling
and Fouling in Membrane Distillation for Desalination
Applications: A Review.” Desalination 356 (January 2015):
294–313.
As Published
http://dx.doi.org/10.1016/j.desal.2014.06.031
Publisher
Elsevier
Version
Author's final manuscript
Accessed
Thu May 26 21:39:27 EDT 2016
Citable Link
http://hdl.handle.net/1721.1/102497
Terms of Use
Creative Commons Attribution-Noncommercial-Share Alike
Detailed Terms
http://creativecommons.org/licenses/by-nc-sa/4.0/
Scaling and fouling In membrane distillation for desalination
applications: A review
David M. Warsingera, Jaichander Swaminathana, Elena Guillen-Burriezab, Hassan A. Arafatb, John
H. Lienhard Va*,
a. Department of Mechanical Engineering, Massachusetts Institute of Technology, 77
Massachusetts Avenue, Cambridge, Massachusetts 02139, United States
b. Institute Center for Water and Environment (iWATER), Department of Chemical and
Environmental Engineering, Masdar Institute of Science and Technology, Abu Dhabi, United
Arab Emirates, , PO Box 54224, Abu Dhabi, UA
* Corresponding author, lienhard@mit.edu
1
Contents
Abstract ......................................................................................................................................................... 4
Nomenclature ............................................................................................................................................... 5
1 Introduction ............................................................................................................................................... 6
2 Types of Fouling in MD............................................................................................................................... 9
2.1 Inorganic Scaling in MD....................................................................................................................... 9
2.1.1 Alkaline Scale in MD ................................................................................................................... 10
2.1.2 Non-Alkaline Scale in MD ........................................................................................................... 15
2.2 Particulate and Colloidal Fouling in MD ............................................................................................ 18
2.3 Biofouling in MD ............................................................................................................................... 21
2.3.1 Bacteria and Biofilms in MD....................................................................................................... 22
2.3.2 Natural organic matter (NOM) in MD ........................................................................................ 25
3 Factors that Influence Scaling in MD ....................................................................................................... 27
3.1 Temperature ..................................................................................................................................... 27
3.2 Dissolved Gases ................................................................................................................................. 28
3.3 Water Source ........................................................................................................................................ 29
4 Scaling and Fouling Effects on MD Operating Parameters ...................................................................... 32
4.1 Wetting and Permeate Water Quality Change ................................................................................. 32
4.2 Permeate Flow Rate Reduction ........................................................................................................ 33
4.3 Increased Temperature and Concentration Polarization ................................................................. 35
4.4 Membrane Damage and Chemical Degradation............................................................................... 37
5 Fouling Mitigation in MD ......................................................................................................................... 41
5.1 Thermal Water Softening .................................................................................................................. 41
5.2 Micro/Nano Filtration Pre-treatment ............................................................................................... 42
5.3 Use of Antiscalants............................................................................................................................ 43
5.4 pH Control of the MD feed ............................................................................................................... 44
5.5 Magnetic Water Treatment .............................................................................................................. 45
5.6 Tailoring MD membrane properties ................................................................................................. 46
5.7 MD module design and operation .................................................................................................... 48
5.7.1 Temperature Polarization .......................................................................................................... 48
5.7.2 Concentration polarization ........................................................................................................ 49
2
5.7.3 Saturation Index, Numerical Modeling ...................................................................................... 50
5.7.4 Effect of Temperature ................................................................................................................ 51
5.7.5 Polarization Abatement: Feed Flow Rate and Bubbling ............................................................ 54
5.7.6 Types of Module ........................................................................................................................ 58
5.7.7 Membrane Cleaning................................................................................................................... 60
6 Trends in Scaling in MD ............................................................................................................................ 63
7 Conclusion ................................................................................................................................................ 63
Acknowledgements..................................................................................................................................... 65
References .................................................................................................................................................. 66
3
Abstract
Membrane distillation (MD) has become an area of rapidly increasing research and development
since the 1990s, providing a potentially cost effective thermally-driven desalination technology when
paired with waste heat, solar thermal or geothermal heat sources. One principal challenge for MD is
scaling and fouling contamination of the membrane, which has gained growing attention in the
literature recently as well. The present paper surveys the published literature on MD membrane fouling.
The goal of this work is to synthesize the key fouling conditions, fouling types, harmful effects, and
mitigation techniques to provide a basis for future technology development. The investigation includes
physical, thermal and flow conditions that affect fouling, types of fouling, mechanisms of fouling, fouling
differences by sources of water, system design, effects of operating parameters, prevention, cleaning,
membrane damage, and future trends. Finally, numerical modeling of the heat and mass transfer
processes has been used to calculate the saturation index at the MD membrane interface and is used to
better understand and explain some of trends reported in literature.
Keywords: Membrane distillation, fouling, scaling, biofouling, membrane cleaning, scale mitigation
4
Nomenclature
membrane distillation coefficient [kg/m2s Pa]
()
ℎ
function of
heat transfer coefficient [W/m2K]
ℎ
enthalpy of vaporization [J/kg]
J
mass flux [kg/m2s]
mass transfer coefficient [m/s]
solubility product constant
pressure [Pa]
heat flux [W/m2]
temperature [°C]
salinity [g/kg]
density [kg/m3]
(⋅)
feed
(⋅)
bulk/free stream
(⋅)
membrane
MED
Multi-Effect Distillation
MD
Membrane Distillation
MSF
Multi-Stage Flash Distillation
RO
Reverse Osmosis
SI
Saturation Index
5
1 Introduction
Membrane distillation (MD) is a promising thermally driven desalination technology still in its
infancy in terms of development and commercial deployment [1, 2]. The technology purifies water
using a hydrophobic membrane, which is permeable to water vapor but which repels liquid water. In
seawater desalination applications of MD, as hot saline feed solution flows over the membrane, the
increased water vapor pressure from the higher temperature drives vapor through the pores
( ≈ 0.2 − 0.4) of the hydrophobic membrane, where it is collected on the permeate side [3]. MD
possesses unique advantages over other desalination technologies, including pressure-driven methods
such as reverse osmosis (RO) and thermally-driven methods such flash distillation. MD is free of the
specialized requirements of high-pressure RO systems, which includes heavy gauge piping, complex
pumps, and maintenance demands [1]. Since MD is not a pressure driven process and only vapor is
allowed to cross through the membrane, MD is more fouling resistant than RO [4] and has a potential
100% rejection of ions and macromolecules. MD can be run at lower temperatures than other thermal
systems making untapped sources of waste heat usable, it requires significantly fewer parts, and can
have a much smaller footprint as result of reduced vapor space [3]. Additionally, recent theoretical and
computational work claims potential multistage DCMD configurations with efficiencies greater than that
of other thermal technologies [5, 6, 7], assuming very large available heat exchanger areas. In practice,
GOR values of practical state of the art MD systems with limited exchange areas are more modest [8].
Summers [9] has subsequently shown that multi-stage vacuum MD is thermodynamically identical to
MSF, indicating that equivalent energy efficiencies can be achieved. The comparative simplicity makes
MD more competitive for small-scale applications such as solar-driven systems for remote areas,
especially in the developing world [3, 10, 11, 12]. However, significant advancements are needed in
membrane technology for MD to reach the theoretical cost competitiveness and develop market share
growth [13]. Fouling in MD is of particular importance, as fouling increases costs of energy consumption,
6
downtime, cleaning, required membrane area, required membrane replacement, and creates problems
with product water contamination from pore wetting [14, 15].
The first patents on MD were granted in the late 1960s, but it wasn’t technologically feasible until
ultrafiltration membranes in the 1980s enabled sufficiently high trans-membrane fluxes [3]. Currently,
most MD work is done in the laboratory, although a number of test beds across the world for small-scale
solar thermal MD have already been deployed, and a few other projects exist [3, 11, 16].
While increased research interest in MD is relatively recent [17], scaling under high temperature
conditions has been a key problem in systems with water heating since the advent of the steam engine.
Research in the area, especially for metal heat exchangers, originated well before 1900 [18]. However,
with respect to thermal efficiency, these studies mainly focus on conductive resistance due to scale
formation, and often do not address the type of transport phenomena that are important in the context
of fluid-membrane systems [18]. A somewhat more relevant area of scaling research is that for RO.
However, RO membranes are not specifically hydrophobic, are virtually non-porous, are comprised of
different materials, and operate at much lower temperatures but much higher pressures. Hence, RO
membranes exhibit significantly different fouling characteristics than MD membranes [3, 14, 18, 19].
Studies focused on scaling in MD largely originated in the 1990s, and since then have become
more numerous [13, 17]. Between 1991 and 2011, sixteen solar-driven MD systems were tested at the
pilot or semi-pilot scale [20]. Limited fouling data from those plants constitute most of what we know
about the fouling potential of MD membranes and the damage they may sustain under actual field
operation conditions. Parallel to those pilot studies, a number of dedicated lab-scale studies were also
conducted to understand fouling in MD. For many years, it was believed that the hydrophobic nature of
the membrane, the maximum pore size and the low feed pressure in the MD process are sufficient to
prevent the feed solution from penetrating the membrane pores (often referred to as pore wetting),
7
and from causing significant scaling on its surface. For example, in 2003, Koschikowski et al. [21] stated
that “the membranes used in MD are tested against fouling and scaling. Chemical feed water pretreatment is not necessary. Intermittent operation of the module is possible. Contrary to RO, there is no
danger of membrane damage if the membrane falls dry.” Indeed, for years it was widely accepted that
MD has this described ability to withstand dryout from intermittent operation. In fact, this is how most
solar-powered MD plants operated, intermittently (shutting down overnight) and allowing the
membranes to fall dry for hours every day [21, 22, 23, 24]. Intermittent operation can also result from
unstable solar conditions or an uneven distribution of flux [24]. In contrast, the present review shows
that while MD membranes are relatively resistant to fouling, they remains vulnerable to it and often
require well engineered designs and operating methods to avoid and mitigate damage or destruction of
the membranes by fouling. These design choices, especially in the case of inorganic scaling, are often
related to maintaining the concentration of ions and the temperature at the membrane interface within
limits where crystallization is not favored. Understanding temperature and concentration polarization
effects (relative reduction in temperature and increase in solute concentration at the membrane
interface compared to the feed bulk, due to the removal of energy and water mass through the
membrane) therefore becomes key. Section 6.7 considers these factors in further detail while
interpreting scaling data available in the literature.
Importantly, current MD membranes are adapted from microfiltration and similar markets, as
yet there are no commercially available membranes specifically made for MD desalination [17]. An aim
of this paper is to summarize differences in membrane properties for desalination from the literature so
as to provide a background for the development of future, specialized membranes. The paper also aims
to better understand fouling mitigation methods, and the effects and risks of different foulants.
8
2 Types of fouling in MD
Fouling is commonly defined as the accumulation of unwanted material on solid surfaces with
an associated detriment of function. The types of fouling that can occur in membrane systems and
therefore potentially found in MD systems can be divided into four categories: inorganic salt scaling or
precipitation fouling, particulate fouling, biological fouling, and chemical membrane degradation [25, 26,
27, 28]. The appropriate mitigation methods vary dramatically for each of these [29, 30]. The causes also
vary strikingly, although particulate fouling can be closely related to the others, as a result of
coagulation. It is therefore most practical to analyze each of these four types separately. After MD was
introduced in the late 60s, the first commercial applications for MD were in the food and semiconductor
industries, not desalination [30, 31]. Since 1985, the number of publications dealing with MD
desalination has increased [31, 32]. As a result, expectations for the types of scaling in MD are often
inferred from other desalination technologies, particularly RO. While both technologies involve mass
transfer through membranes, significant differences related to fouling exist, notably the significantly
higher operating temperatures of MD, as well as the hydrophobic properties of MD membranes, the
presence of temperature gradients in MD, and the larger pore sizes in MD [3, 33]. Also, MD lacks the
high pressures of RO which are generally believed to aid the formation of compacted cake scales.
2.1 Inorganic scaling in MD
Study of inorganic scaling dominates the fouling literature for MD. Inorganic scaling, or simply
scaling, in RO and MD generally falls into one of three categories: alkaline, non-alkaline, and uncharged
molecule scale [14, 34]. Alkaline salts, or basic salts, have a tendency to make a solution more basic
when added through hydrolysis, forming hydroxide ions. Generally, acidifying solutions below pH 7
decreases the tendency for many alkaline salts like calcium carbonate to scale or precipitate [35]. Nonalkaline salts include most other charged ions that dissolve in water without pH-raising tendencies [36].
9
Uncharged molecules that may scale, such as silica, are generally less soluble than salts because the
charge on salts allows highly polar water molecules to break up and dissolve the salts. Uncharged scale
can be considered with particulate scale [37], and is done so in this work.
2.1.1 Alkaline scale in MD
Calcium carbonate is perhaps the most common scale in thermal desalination systems, often
limiting operating conditions in brackish, groundwater, and seawater desalination [19, 38]. In thermal
and RO desalination, CaCO3 scale is regarded as a pervasive scale and among the first to reach
supersaturated conditions in many feed solutions [38]. Calcium carbonate scale often forms after the
breakdown of bicarbonate, HCO3-, as shown in the equation below:
!"# + 2% &'( → ! &' + &" + %" &
For typical MD conditions, the breakdown of bicarbonate plays the dominant role in calcium
carbonate precipitation [39]. However, the carbonate equilibrium and scaling is more complex than this
simple equation implies. Increased pH and higher carbonate concentration are strongly associated with
calcium carbonate scale [40]. The solubility of CaCO3 changes dramatically with the concentration of
CO2, and may decrease at higher temperatures as CO2 comes out of solution, which raises the pH [19,
41]. Adding to this effect, CaCO3 has inverse solubility, so high temperatures will decrease its solubility
irrespective of CO2 concentration [39]. According to Shams El Din, in typical thermal desalination
systems such as MSF, a temperature of 37 oC can be considered as the minimum temperature for the
formation of CaCO3 from mildly concentrated ocean water [42], which is well below the typical
operation temperatures of MD (60-80oC). However, the alkaline scaling process is strongly dependent on
many factors such as heat transfer rate, brine concentration, residence time, flow conditions, etc. [19].
10
In general, bicarbonate dissociates more readily with increased temperatures. Notably, at higher
temperatures carbonate has a tendency to hydrolyze into carbon dioxide [43], as follows:
&'"( + %" & → 2&% ( + &" This reaction makes the solution more basic, which influences the solubility of other scales,
notably making Mg(OH)2 more likely to precipitate. Carbon dioxide gas may come out of solution; this
process is related to thermal water softening that can be used as a pretreatment strategy and has been
discussed in the fouling mitigation (Section 5).
Calcium carbonate precipitates can take six different forms. Three anhydrous crystalline
polymorphic forms may occur, known as calcite, aragonite and vaterite. These forms, all CaCO3, differ in
crystal morphology, color, hardness, and refractive index [44]. Three hydrated forms occur as well:
amorphous calcium carbonate (ACC), calcium carbonate monohydrate (MCC), and calcium carbonate
hexahydrate (CCH) [19, 45, 46, 47, 48]. Calcite is the most stable form found in MD system operation,
but vaterite is common as well [49, 50]. In MD experiments, it was found that calcite formation was
promoted by laminar flow [49]. Aragonite has been observed for MD as well [51], but it is relatively rare.
The form of calcium carbonate is highly dependent on temperature, as seen in Figure 1 [39].
11
Fig. 1. Solubility of several forms of calcium carbonate in water as a function of temperature [52].
Importantly, calcium carbonate scale in MD is usually only observed at relatively high saturation
indices (SI). However, the presence of a microporous membrane substantially reduces induction times in
CaCO3 precipitation, causing an increase in nucleation rate regardless of the SI index [53, 54]. With
hollow fiber membranes, Gryta observed that for saturation indexes between 5 and 20
(supersaturated), which he considered low levels, the induction period for CaCO3 scale exceeded 30 min
for tests ranging from 20°C to 100°C [39, 55]. Fei et al. [50] found that the system required very high SI
values for precipitation of CaCO3 in MD, recording concentration 32 times higher than saturation
concentration to initiate scaling for calcite [50]. Fouling of CaCO3 may be highly variable. As solubility
decreases significantly with temperature, Gryta recommended a feed temperature below 80°C to avoid
calcium carbonate scale for experiments with lake water [39]. Although more common at higher
temperatures due to reduced solubility, calcium carbonate scale has been an issue for MD even at low
temperatures, such as 40°C in a study on untreated tap water with stacked membrane modules [56].
12
Calcium carbonate fouling can even occur at ambient temperatures such as in reverse osmosis at
sufficiently high concentrations [38].
Fig. 2. SEM image of calcium carbonate scale on a polypropylene Accurel PP S6/2 membrane surface
with tap water feed for long duration DCMD experiments. Experiment performed with 86oC feed and
20oC distillate [57].
Studies of calcium carbonate scale in MD have consistently found that pure calcium carbonate
scale causes significant flux decline, and that it is generally nonporous in nature [39, 58]. However, it has
also been found that the feed flow rate can modify the morphology of the carbonate deposits. Fig. 2
illustrates CaCO3 crystal scale formation under small Reynolds number flow in hollow fiber membranes.
In general, lower flow rates promote the growth of bigger crystals and more compact scaling layers,
while higher flow rates reduce the size of the crystals and create comparatively more porous layers [39].
The flux decline varies widely, from zero or near zero to overall declines as high as 66% [59].
Furthermore, many experiments have found that calcium carbonate penetrates and scales in pores [58].
The scaling of calcium carbonate frequently causes wetting and thus results in contamination of the
permeate by the feed [2, 49]. Long term performance studies, on the order of thousands of hours, found
that calcium carbonate was consistently associated with wetted membranes, while deionized water
experienced no such wetting [57]. Discussion of membrane damage from calcium carbonate scale can
13
be found in the following sections, and results of calcium carbonate experiments, including flux decline
or fouling rate, are detailed in Table 1.
Notably, some studies have pointed out that the content of impurities in the CaCO3 solutions
used as feed can definitely play a role in flux decay. For example, He et al. [50] reported almost
negligible impact on permeate flux using a very pure CaCO3 solution. Fast homogeneous precipitation in
the bulk solution and transport of CO2 across the membrane explained the negligible membrane scaling,
which is in good agreement with the results presented by Nghiem et al. [60].
Co-precipitation of foulants is common in desalination systems, and complicates the prediction
of scaling behavior. Calcium carbonate co-precipitation has been frequently observed in MD systems,
especially with calcium sulfate [58]. However, systematic studies of co-precipitation are lacking [25, 61].
Gryta found that CaCO3 co-precipitation with CaSO4 resulted in bimodal crystal size distribution, using
scanning electron microscopy with energy dispersive X-ray spectrometers (SEM-EDS) post mortem
analysis. Importantly, co-precipitation with CaSO4 was found to weaken the negative effects of CaCO3,
including reducing wetting and reducing membrane damage [58]. However, in a different paper, Fei
found that co-precipitation of CaCO3 and CaSO4 caused an increased flux decline relative to CaCO3 alone
[50]. Studies on CaCO3 and CaSO4 precipitation kinetics have reported that an increased carbonate level
may make CaSO4 scale more tenacious and fine, but slows the rate of CaSO4 precipitation [61]; this may
explain the seemingly conflicting results in literature on the co-precipitation of these two compounds.
Curcio et al. [25] analyzed the fouling of CaCO3 in the presence of humic acid (HA), with
synthetic seawater concentrated 4 to 6 times. They found that the presence of other ions, including
magnesium, sodium, sulfate, and HA, all inhibited the precipitation of CaCO3 [25], which agrees with
other studies [25, 62]. With 2 mg/L of HA present in hollow fiber membrane MD, the induction time of
CaCO3 was increased from 16 to 30 seconds [25]. It was found that HA increases the interfacial energy of
14
vaterite by 7%, from 45 to 48 mJ/m2 [25, 62]. Other authors have shown that different humic
substances, such as humin or fulvic acid, have dissimilar degrees of inhibition on calcite growth [25, 63].
Scaling literature predicts that the interactive effects of mixed salt solutions are significant and
can alter the thermodynamics of precipitation [64]. Gryta found that when CaCO3 co-precipitated with
iron oxides, the scale was porous, and the flux reduction was not very high [26].
Several magnesium scales may also be a concern in MD for feed solutions with high levels of
Mg2+. Magnesium hydroxide is another commonly observed alkaline scale in desalination applications,
especially in groundwater, albeit not nearly as pervasive as calcium carbonate scale. Like CaCO3, it
exhibits inverse solubility with temperature, increasing its scaling propensity in MD [39]. Gryta tested
lake-derived tap water with an Mg concentration of 15 mg/L in a Direct Contact Membrane Distillation
(DCMD) system, and found that Mg(OH)2 scaling occurred at above 348 K.
2.1.2 Non-alkaline scale in MD
Calcium sulfate is one of the most common non-alkaline scales that occur in membrane systems
[19]. In thermal desalination, CaSO4 scale is regarded as a tenacious and very adherent scale [58], and it
has behaved as such in MD processes as well [58]. Cleaning calcium sulfate is relatively difficult
compared to alkaline scales, so modifying operating conditions to avoid such scale is the most common
method of mitigation [19]. Calcium sulfate scale may occur in one of two hydrate forms, the dihydrate
CaSO4·2H2O (gypsum) and the hemihydrate CaSO4·0.5H2O (Plaster of Paris), or as an anhydrite, CaSO4.
The form precipitated depends strongly on temperature, with gypsum common around 20°C [19, 65],
and the anhydrite form more common at higher temperatures. Calcium sulfate solubility peaks around
40 °C [58], but does not vary dramatically across typical MD operating conditions. A study by Gryta on
calcium sulfate in MD found that the concentration of sulfate ions should not exceed 600 mg/L, but up
15
to 800 mg/L can be tolerated if bulk removal is available [58]. The study, specifically focused on CaSO4
scaling in MD, found that membrane damage caused wetting and leaking and that it prevented further
use of the membrane [58]. Gypsum scale was found to scale and penetrate the membrane pores. SEM
images revealed needle-like gypsum crystals in typical orthorhombic and hexagonal prismatic needle.
The crystals were tightly packed and tended to grow outward from initiated sites. As a consequence,
exponentially worsening flux decline was observed, roughly experiencing a 29% decline over 13 hours
[58]. Studies suggest that a supersaturated condition alone is not enough to start the crystallization of
CaSO4 on the membrane surface. Sufficiently long induction times (i.e., 53, 43, and 30 h for feed
concentrations of 500, 1000, and 2000 mg/L of CaSO4, respectively) [60] are also a requisite [60]. This
long induction time suggests a strategy to control the CaSO4 membrane scaling.
Calcium phosphate, another potential scale, is a non-alkaline scale that has frequently occurred
in wastewater treatment and in RO membranes [19, 66, 67, 68]. Phosphate often exists in water
supplies as phosphoric acid, which is relatively weak and which dissociates through several stages;
significant concentrations of phosphate ion do not occur until the pH becomes relatively basic.
Therefore, maintaining a low pH is an effective method to avoid phosphate scale [19]. It is often treated
with use of dispersants in nanoparticle form in the feed as well [19]. However, calcium phosphate scale
has not been found in the MD desalination literature. A most likely potential risk of calcium phosphate
scaling arises when phosphates additives are used as antiscalants. These additives prevent calcium
carbonate precipitation by sequestering Ca2+. However, under relatively high temperatures and neutral
pH (MD conditions), the rate of polyphosphate hydrolysis increases [69] decreasing the scale inhibition
efficiency and creating a potential for calcium phosphate scaling [70].
MD experiments on non-alkaline magnesium scale have been performed as well. Tung-Weng et
al. [71] found that MgCl2 and MgSO4 scale significantly more on polytetrafluoroethylene (PTFE)
16
membranes compared to polyvinylidene fluoride (PVDF) membranes in a flat sheet module. They further
report flux rate reduction of about a 86% of the initial value with the addition of 0.1% of either MgCl2 or
MgSO4 to a 4.4% NaCl solution. In contrast, increasing the NaCl concentration to a 10% reduced the flux
rate only to a 96% of the initial value, suggesting that the former registered decrease is a result of the
Mg salts precipitating at the membrane surface rather than just a result of increased concentration and
concentration polarization effect.
While not a common scale for most MD installations, sodium chloride, as the principal
constituent of most desalination feeds, has been very widely used in MD literature including scaling
studies. Sodium chloride, a non-alkaline scale, is characterized by a very high solubility and lengthy
induction times. In scaling experiments by Tung-Wen Cheng et al. [71] under DCMD conditions at 50°C,
increasing NaCl concentrations from 4.5% to 10% by weight only resulted in a 3-4% flux reduction, an
expected level because of the decreased mole fraction of water at the membrane surface and not
indicative of scaling-induced flux reduction, confirmed by the SEM micrographs which showed small
levels of crystallization [71]. A MD paper by Fei He operating at 10% wt. NaCl reported similar results [7].
By contrast, extreme concentrations of NaCl, roughly 26-27.5 wt%, resulted in significant fouling [72].
After about 26% NaCl, the feed concentration seemed to asymptote while the flux dropped
dramatically, indicating the onset of significant scaling. In the experiment, for roughly the first 250 min.,
26% NaCl gave very good agreement to theory, indicating that even at this high concentration, the flux
only dropped due to reduced vapor pressure, not scaling [73]. The extremely high solubility of NaCl
relative to other salts and consideration of available sources of water indicates that virtually no natural
source of water for desalination would have NaCl fouling as a concern. However, in cases of drying out
membranes, NaCl will be among the salts to easily form on the membrane surface, as it will be discussed
later. Other studies have shown that high concentrations of salts, exceeding saturation, can also cause
wetting [58, 74].
17
2.2 Particulate and colloidal fouling in MD
Particulate and colloidal fouling risk is common in many feed water solutions. Larger particles
can often be addressed with modern filtration technology (i.e., UF, MF, NF), but smaller particles can be
an issue in fouling. Notably, particles vary greatly by water source, and, in the case of surface water, by
season. However, in many MD desalination pilots, the use of cartridge filters or screens is widespread
and proves to be effective for particulate matter [75, 76, 77]. Compared to ocean water; lake, ground,
and especially river water are more likely to have particulate fouling concerns. Particles and colloids
include clay, silt, particulate humic substances, debris, and silica [78].
Silica is particularly notable because its small size makes it harder to remove with pretreatment
stems such as microfiltration. Silica is generally found in water supplies in three forms: colloidal silica,
particulate silica, and dissolved silica (or monosilicic acid). The latter can cause severe fouling in RO and
FO systems when supersaturation is reached and the silica starts polymerizing on the membranes [19].
PH can also play a role in the ionization and polymerization of silica; nevertheless at most natural pH
levels (including that of the SW, around 8.5) silica is relatively unionized, lowering the risk of scaling [79].
However, silica solubility increases with temperature and should be much less of a problem in thermal
systems such as MD. In an MD experiment with hollow fiber membranes and tap water, Karakulski et al.
[59] found precipitation of silica compounds on the membrane [59]. The silica clogged capillary
membrane inlets, causing a gradual decline of the module efficiency. The flux declined by 30% after
1100 hours of operation, mostly during the first 200-300 hours. Removal of the foulant with acidification
combined with drying the membranes restored the initial flow rates only briefly [59]. This happened
despite nanofiltration of the feed upstream of the MD membranes [59]. SEM-EDS analysis indicated that
the deposit consisted of silicon, with small amounts of iron, calcium, zinc, and chlorine. Unlike
conventional fouling, which coats the surface and blocks the pores, the SEM analysis indicates that the
18
decline was not from a deposit layer, but from clogging membrane capillaries. The clogging reduced the
feed flow rate, thus increasing temperature and concentration polarization, which reduced the module
flux. The fibrous structure of the deposits blocked the foulants from further entry into the membranes
capillary, stabilizing the flux [59]. This indicates that silica scale may be of significantly more concern in
hollow fiber capillary MD systems than in flat sheet membranes. Silica fouling, while not causing a flux
decline as fast as calcium carbonate, is a concern because it is difficult to clean. Acids that are commonly
used to break down crystalline scale are not very effective on silica, which is uncharged [59]. When the
feed has significant silica present, the authors recommend avoiding hollow fiber membranes with feed
flow inside the capillaries.
An important and typically particulate foulant investigated in MD is iron oxide. Iron oxide fouling
may consist of a variety of compounds, including iron oxides, iron hydroxides, and iron oxide-hydroxides
[26]. These compounds are usually crystalline, and also may consist of hydrated forms. Iron oxide scale
is not anticipated to be present in typical feed waters, but it is a risk of scaling due to the high
propensity to rust on steel and even stainless steel parts in distillation systems. Corrosion fouling cannot
only cause clogging problems, but also cause membrane damage by surface erosion (corroded flakes
and chunks in motion through the narrow flow passages). Gryta found significant iron oxide fouling
unexpectedly in a study on MD for treating effluents from regeneration of an ion exchange system in a
water treatment plant [26]. However, the study found that iron oxide deposits did not significantly
affect flux, undergoing only an 8% decline in permeate flux over 20 hours of operation. It was inferred
that the iron scale was relatively porous. The composition of the iron oxides foulants was determined by
x-ray diffraction, including maghemite, lepidocrocite, akaganéite, and hematite. It was found that the
“iron oxides, hydroxides, and oxide-hydroxide” scales were crystalline [26]. These oxides exhibited high
tendency to accumulate both on the membrane surface and within membrane pores.
19
The expected wetted corrosion reaction that causes iron oxide scale, also known as
electrochemical corrosion, is as follows [26, 80]:
4Fe + 2H" O + 3O" = 2Fe" O' ⋅ H" O
Corrosion reactions vary, but consist of oxidized forms of the metal and particles of positive, negative, or
zero charge. Reactions also vary by oxygen content present [26]. In the various corrosion reactions, at
least one of the products will be an oxidized metal, a metal cation, metal anion, or uncharged solid
compound [26]. Oxidized metals may be Fe2O3 , FeO, or Fe3O4; iron metal cations are Fe2+ or Fe3+; metal
anions include HFeO2- and FeO42-; and uncharged solid compounds include Fe(OH)3, Fe3O4, and
Fe2O3*H2O [26]. The oxidation reactions are complex, and are affected by conditions in the feed,
including salinity, feed composition, and oxygen content. Under oxygen limited conditions, black
magnetite Fe3O4 is often formed. When other salts are present, such as Cl- or SO4-, they may be
incorporated into iron oxides or hydroxides. Hydrolysis of Fe3+ ions may occur in basic conditions, from
heating, or dilution of a salt with the ion. Hydrolysis may form hexaaquocation Fe3(H2O)63+, and its H2O
ligands again experience hydrolysis, creating FeOH or Fe2O3. The oxides present and their crystalline
structure generally vary by conditions of formation, including temperature, other ions present, and pH.
The presence of water, high or low pH, and other dissolved ions are conditions existing in MD that
encourage corrosion of steel elements [26].
Additional Iron Oxide fouling in Gryta’s study occurred as a consequence of acid
cleaning (HCl) of the feed side. The volatile acid was capable of getting through the membrane to a small
degree as gaseous HCl, acidizing the permeate and causing oxidation of the stainless steel elements on
the permeate side of the system. Even concentrations of less than 50 g/L HCl can lead to significant
reaction with the stainless steel elements [26]. Therefore, Gryta recommends using acid resistant highgrade steel or plastic as materials for MD systems [26, 81]. However, it is important to note that Gryta
20
used very acidic conditions to clean the module, 18% and 36% HCl, which are relatively high compared
to 3% or 5% HCl used in other experiments [26]. Therefore, acidic permeate may not be guaranteed to
be a concern. The acid wash trials from the study are discussed in Section 5.7.7. The fouling layer was
observed with SEM-EDS, as seen in Fig. 3, and small amounts of Cu, Zn, Ca, P, Al, Mg, Na, S, Cl and Si
were also observed in the membrane.
Fig. 3. SEM micrograph of the polypropylene membrane surface covered with iron oxides after cleaning
with 18 wt.% HCl failed to remove them, with a 353 K feed and a 293 K distillate temperature [26].
Compared to other scale forms in MD, iron oxide can be judged as unlikely to occur with proper
system design. It is relatively less harmful to permeate flux but still a major cause of wettability and very
difficult to remove. The lower operating pressure typical of the MD process makes it possible to use
plastic components, which can potentially eliminate most iron from the system.
2.3 Biofouling in MD
Biofouling of hydrophobic membranes applicable to MD has been a key research interest in the
food, beverage, and wastewater industries [27, 82]. Many of the membranes used in MD originated in
21
these industries. The majority of these studies focus on very high chemical oxygen demand (COD)
effluents, although recently MD biofouling studies have extended to MD for desalination [27]. The high
COD content comes from using MD or similar systems for animal products processing, fermentation, and
other processes [27]. For the sake of brevity, this review focuses only on biofouling relevant to clean
water production.
Biofouling relevant to MD desalination includes bacteria, fungi, and biofilm studies. Biofouling is
pervasive in most waters [78] and has been a critical issue for RO membranes [83] and it is likely to be a
concern in practical MD systems. However, the operating conditions of the MD process, especially the
high temperatures and salinity, can restrict to a great extent the microbial growth in MD installations
[27]. As a consequence, the problems caused by biofouling in membrane processes including NF, UF or
RO should not occur in such a high degree in MD systems. [84]. However, organic fouling can play a
more important role in membrane wetting in MD.
2.3.1 Bacteria and biofilms in MD
Bacteria and microorganisms are pervasive in water systems. While chlorination is effective in
killing bacteria, it can be damaging to many common MD membrane materials [78]. Bacteria can be very
difficult to remove from membranes, as they excrete an extracellular polymer substance (EPS) to adhere
to the surface [78]. In typical biofilm formation, bacteria colonize and excrete EPS, and then organic
compounds accumulate in the film. These compounds are typically composed of polysaccharides,
proteins, lipids, humic substances, nucleic acids and aromatic amino acids, and often contain trapped
particles and absorbed substances [27]. These biofilms are typically 75-95% water and are relatively
porous compared to alkaline scale [78].
Biofouling impairs MD process through wetting and pore blocking. Additionally, the relatively
porous biofouling layers reduce diffusion and create a hydrodynamically stagnant layer of water at the
22
feed side [27]. Biofilms are mostly constituted by a hydrated EPS matrix which makes diffusion the main
mass transport mechanism. Water diffusion coefficient in biofilms has been estimated to be 15% lower
than that in bulk water [85], conferring biofilms an extra mass transfer resistance and increasing
temperature and concentration polarization effect. As a consequence, they can hinder convective heat
transfer to the membrane while favoring diffusion and conduction. The thermal conductivity of biofilms
has been estimated to be around 0.57–0.71 W/mK (close to that of water), almost 75% less than that of
inorganic scale (i.e. CaCO3 and CaSO4) [86].
In Krivorot et al.’s experiments on hollow fiber MD with ocean water and a high biological load
(1×108 CFU/ml) at 40°C, permeate flux declined by 34% over 19 days [27]. However, minimal flux decline
was detected in samples with normal biological loads. The high biological load was attributed to local
wastewater spills to the sea. In the high biological load sample, a conditioning biofilm was formed in as
little as 4 hours, and a recognizable biofilm was apparent after 28 hours. Over the 19 day experiment, all
samples showed a biofilm. Temperature cycling to 70°C was found to reduce the biofouling behavior. In
general, processes with a top brine temperature of 70°C minimized any biological presence [27].
In a study done by Gryta [84], MD was performed on a bioreactor with saline wastewater that
contained yeast, Pseudomonas and Streptococcus Faccalis bacterium, and the fungi Penicillium and
Aspergillus. The DCMD hollow fiber membranes failed to prevent Streptococcus bacteria from entering
the distillate. With a temperature of 90°C and salts concentrated up to 300,000 ppm, no bacteria was
detected at the membrane surface, indicating these conditions prevented bacterial growth. However,
when the temperature was decreased to 80°C, bacteria and fungi were detected at the membrane
(Fig.4).
23
Fig. 4. Left: SEM image of Aspergillus fungi on MD membrane surface. Right: SEM image of
Streptococcus faecalis bacteria, 65°C feed solution, at roughly 5 times the concentration of seawater
[84].
Notably, while the anaerobic bacteria Streptococcus grew on the membrane, no aerobic bacteria
were observed on the membrane, despite being present in the original feed. The same feed caused
significant fouling in MSF and RO, but tolerable and unproblematic fouling in MD. Elevated temperature,
significant salt concentrations, and low pH values all can hinder bacterial growth. Similar MD studies by
Meindersma et al. using an AGMD system and pond water as feed, reported a flux decline after 800 h
from biofouling. No organism break through was noted however [87]. In this same study, original flux
was almost restored by reversing the direction of the flow.
Biofilms, especially in seawater settings, often contain more microorganisms in addition to
bacteria. Although direct MD experiments with marine microorganisms couldn’t be found in the
literature, highly relevant superhydrophobic materials similar to MD membrane materials have been
examined in seawater conditions. A study by Zhang [88] compared ocean fouling in submerged
hydrophobic and superhydrophobic surfaces over 6 months. Polysiloxane and PTFE surfaces were
examined. The results showed that the hydrophobic surface exhibited fouling within a day, but the
superhydrophobic surface (contact angle 169°) resisted fouling for about three weeks. However, after 2
months, both surfaces were heavily fouled and wetted with 10-20% macroalgae, 5-10% barnacles, and
24
50-60% bryozoans, although they resisted the tubeworms, ascidians, and to some extent algae that
covered the control plate. The loss of biofouling resistance was attributed largely to air bubble loss
through the membrane, as the air dissolves into the surrounding water [88]. These results can be
extended to MD surfaces, where seawater is passed on the feed side over a hydrophobic membrane.
Specific conditions for avoiding biofouling cannot be determined presently from the literature
because biofouling depends on many factors such as: salt concentration, feed composition, residence
time, pre-treatment, bacterium present, operating temperatures, membrane type and cleaning
frequency [27].
2.3.2 Natural organic matter (NOM) in MD
NOM compounds at risk for MD are especially prevalent in wastewater and certain lake and
ocean water samples. NOM includes various constituents such as proteins, amino sugars,
polysaccharides, polyhydroxy-aromatics, and humic substances [89]. These compounds are often
present where traditional biofouling is a concern, as they often provide sustenance to bacteria and may
originate from similar processes. The fouling caused by NOM can affect both the permeability and the
dissolved solids rejection of the membranes. Membrane fouling in the presence of organic compounds is
affected by ionic strength, pH, ions present, membrane surface structure and chemistry, molecular
weight, polarity, permeate flux, and hydrodynamic and operating conditions [34]. In general, hydrophilic
surfaces are less susceptible to organic fouling [90, 91] but MD employs hydrophobic surfaces, making
organic fouling a concern.
The principal NOM foulant is HA, a general term for complex mixtures of organic acids with
carboxyl and phenolate groups [92]. HA is produced by biodegradation of organic matter, and it gives
many swamps and rivers a characteristic yellow brown color [36]. HA are complex and vary greatly;
molecular weights typically occur between 700 and 200,000 Daltons [93], and may even be above or
25
below this range. HA systems are thus typically measured with averages. For instance, the average HA
particle in a solution may act as an acid with two and sometimes three free hydrogen atoms [92]. HA
production in rivers and lakes is often seasonal, with large quantities produced annually from deciduous
tree leaf decomposition. Importantly, HA may readily nourish bacteria, and thus may instigate significant
bacterial fouling [27]. HA fouling in MD may vary based on feed composition, membrane
hydrophobicity, temperature, membrane pore characteristics, and pH [94, 95]. For example, the
addition of multivalent cations increases the electrolyte or ionic strength of the feed water and can
favor the aggregation of the HA and therefore the fouling [96]. Divalent ions, including Ca2+, act as
binding agents to the carboxyl functional groups reducing the charge and the electrostatic repulsion
between macromolecules and encouraging particulate precipitation [25, 97]. HA at 100mg/L
concentration in a MD study with added CaCl2 at 3.775 mM caused significant flux reduction of 40% on
flat-sheet membranes [29]. This solution treated with MD produced a thick fouling layer, blocked pores,
and increased heat transfer resistance [29]. In more acidic conditions, HA dissociates less because fewer
Ca2+ ions are available [29, 98]. However, HA can also affect other types of scaling and has significant
scaling inhibition effects on calcium carbonate (i.e. inhibit heterogeneous nucleation and increases
induction time) as explained earlier [25].
Other metal ions (i.e., Fe3+, Mg2+, Al3+, etc.) can also contribute to NOM fouling in a similar way:
increasing the ionic strength and causing metal ion-induced aggregation [99]. In a study with UF hollow
fiber polyethersulfone (PES) membranes, the presence of Fe3+ in the HA feed solution reduced the flux
to one fifth of the original rate in one hour. The effects of Ca2+ and Mg2+ were similar but not as
significant [100]. In this same study, the authors found that the use of EDTA as a chelating agent
inhibited the crosslinking of HA induced by the presence of the metal ions and reduced the fouling.
26
HA deposits are typically loosely packed and porous and traditionally in UF and MF systems they
are effectively eliminated through backwashing. In MD systems, they can be cleaned rather effectively
with basic solutions. Srisurichan [29] found that HA fouling was easily removed with a 0.1 M NaOH
solution, while still achieving full recovery of permeate flux. Alternating temperature changes, in one
case between 25 °C and 35 °C, was found to clean HA as well, resulting in a flux recovery of 98.2% with
proven repeatability [97].
3 Factors that influence scaling in MD
3.1 Temperature
Temperature is among the most dominant factors related to scaling and fouling of MD
membranes. In particular, the solubility and crystal formation of salts vary widely over the temperature
range relevant to the MD systems. Importantly, the solubility of individual salts may be positively or
negatively correlated with temperature. For example, the solubility of sodium chloride increases with
temperature, whereas those of calcium carbonate, magnesium hydroxide, and calcium phosphate
decrease with temperature. This negative correlation of solubility with temperature is typical for alkaline
salts, which depend on the breakdown of water into hydrogen and hydroxide in order to form scale;
such dissociation increases at higher temperatures [36]. Salts such as calcium sulfate and calcium
carbonate that exhibit inverse solubility are also often the closest to being saturated in desalination feed
solutions (calcium sulfate concentration is higher in the case of seawater as a feed while calcium
carbonate concentration is higher in ground water sources). Generally, for common feed solutions,
increased temperature causes increased risk of scaling. Higher temperatures also reduce induction
periods for some salts [101].
27
Temperature can have a significant effect on biofouling due to microorganisms’ lack of tolerance
for high temperature and also because of thermal effects on organic compounds. According to M.
Krivorot et al.’s experiments with hollow fiber membranes, at temperatures above 60 °C, most
environmental organisms will not function and hence not grow on MD membranes [27]. Temperature
increase causes the decomposition of HA and other biological compounds. In fact, for RO membrane
systems, temperature increase may be used as an effective cleaning method for HA [97]. However, for
MD at higher temperatures, permeate flux increases: this may lead to higher concentration of organic
compounds at the membrane interface due to the concentration polarization effect. Srisurichan [29]
found that at higher temperatures, flux decline was greater for solutions containing HA, observing a 16%
decline at 50°C and a 43% decline at 70°C. Severe protein fouling was observed at temperatures higher
than 20–38°C for aqueous solutions containing organic compounds at representative concentrations
(i.e., wastewater, NOM, bovine serum albumin, etc.) [30, 84, 102] but it was practically absent at lower
temperatures [103]. Notably, hydrophobic surfaces show an especially high tendency to get fouled by
proteins [104], making MD membranes problematic for waters containing proteins, amino sugars or
polysaccharides [105].
3.2 Dissolved gases
Dissolved gases are present in almost all feed waters of interest in desalination [36], and these
gases may have some limited effects on scaling and fouling in MD. Gases dissolved in feed water, as well
as those resulting from chemical processes such as the breakdown of bicarbonates, travel into the
membrane along with the water vapor, providing an additional diffusive resistance for the water vapor
[39, 72].
The effect of dissolved gases on fouling is indirect and small; dissolved gases impede the
permeate flow process, reducing concentration polarization and scaling. This occurs because dissolved
28
gases in the feed stream may flow into the membrane pores, providing mass transfer resistance to
water vapor, and may also contribute to mass transfer resistance in the air gap after the membrane
(depending on the system configuration). The effect is to reduce the condensation heat transfer rate,
possibly making the system mass transfer limited on the air-side, thus reducing the overall vapor flux
[16]. On the other hand, the absence of dissolved gases can increase membrane wettability by removing
the air trapped in the membrane pores, which was experimentally confirmed by Schofield et al. [106].
The presence of these gases in the membrane can act as a barrier to fouling. Therefore, reducing
dissolved gases by deaeration or other means may be expected to increase fouling potential [16].
Dissolved gases, especially carbon dioxide, may alter the pH of the solution, affecting the
solubility of various salts as described previously. Dissolved CO2, common in many feed waters and often
produced by breakdown of calcium carbonate in ground water, may acidify the water by the creation
and dissolution of carbonic acid as follows [28]:
&" (!) + %" & ↔ %" &' ↔ % &'( + % # ↔ &'"( + 2% #
So despite increasing carbonate concentrations, typically increased dissolution of CO2 reduces
scaling by the associated pH decrease. This result may differ depending on temperatures,
concentrations, and pH, as discussed in Section 5.1 (on thermal softening). Dissolved gases or lack of
them may significantly affect biofouling, as the presence of dissolved oxygen supports aerobic bacteria
and microorganism fouling. Therefore, deaeration of oxygen may be used to inhibit microbial growth.
3.3 Water source
As seen in previous sections, fouling likelihood and type of fouling in MD depend on the salts
and other foulants present in the feed water and thus are highly dependent on the water source.
29
Generally, specific sources have fairly consistent conditions and thus consistent expectations for fouling,
although surface waters’ quality and algae blooms may be seasonal.
Possible water sources for desalination include lake, river, ocean, ground waters as well as
industrial waste water. Generally calcium carbonate saturation is a significant concern in most water
sources relative to other salts. However, water sources are variable enough that other insights on types
of fouling susceptibility may not be comprehensive. Compared to ocean water, lake and river water are
typically characterized by high silica content, biological compounds, suspended solids, and calcium
concentrations, but vary widely between different rivers and lakes [107]. Due to the low salinity of these
waters, MD use is unlikely, but electrodialysis desalination may be used if slightly reducing salt
concentrations is desirable. Ocean waters often have relatively high scaling potential for calcium sulfate
compared to other surface waters, as well as calcium carbonate, possible biological compounds and
organisms, and a significant dry-out concern from very high levels of sodium chloride [36]. Groundwater
sources are perhaps the most variable. Groundwater often has high levels of salts compared to nonocean surface water [108], and may be especially rich in calcium, bicarbonates, magnesium and sulfate.
Groundwater is also commonly rich in iron (reduced Fe2+) that can oxidize in contact with air and form
iron hydroxides (nearly insoluble in water), producing heavy fouling. Due to its variability, groundwater
may cause some of the worst scaling and flux reduction seen in desalination systems. Finally, the
composition of industrial wastewater varies depending on the source, but can be extremely saline with a
variety of dissolved metals (as in, e.g., produced water from hydraulic fracturing operations). The
compositional variability of these waters makes it difficult to single out problem-causing compounds in a
general fashion.
Gryta investigated MD fouling for water from river, lake, and groundwater sources for hollow
fiber MD [72]. In that study, within 20 hours of MD testing, river tap water experienced the largest flux
30
decline, lake tap water experienced the smallest flux decline, and groundwater had a flux decline in
between. The flux decline was largely caused by calcium carbonate, with bicarbonate ion concentrations
being the limiting factor for causing scale.
In contrast, seawater has very consistent constituents, generally has greatly more sodium
chloride than other bodies of water, very high salt content overall, and lower concentrations of many
other ions than some lake or river waters. Due to typically low concentrations of magnesium and organic
HA, seawater may be less prone to fouling by these components than other sources of water. Like many
tap water sources, calcium carbonate is a significant component of expected scale for ocean water, but
is even more of an issue in ground and lake water. Calcium sulfate is also a concern for scaling in
seawater [109]. Ocean water is susceptible to a variety of biological fouling types, including algae and
microorganisms that may differ significantly from inland waters [27]. Curcio et al. found scaling of CaCO3
in DCMD of seawater at concentration factors of 4-6 and 40°C [25]. The consistency of seawater salts
and literature studies indicate that calcium sulfate, calcium carbonate, particulate fouling, and
biofouling are the expected fouling concerns in seawater.
Wastewater treatment has also been attempted through MD, and fouling studies can even be
found in literature. Wastewater contains numerous fouling compounds that may affect the flow
significantly, especially biological compounds. In a study with hollow fiber MD membranes, a rapid
decline in permeate flux was observed in solutions containing HCO3-, expected from CaCO3 fouling [110].
The presence of bacterium S. faccalis was detected as well, but temperatures of 85°C prevented
bacterial growth. Silica was detected as a deposit as well. Wastewater constituents vary dramatically by
the source, so inorganic scale types may be hard to predict, although organic fouling of various kinds is
common in wastewater.
31
4 Scaling and fouling effects on MD operating parameters
4.1 Wetting and permeate water quality change
An important requirement for the MD process to perform well is that the membranes have to
remain hydrophobic, thus allowing only vapor and not liquid water to pass through. Wetting refers to
the process whereby the membrane starts allowing liquid water to flow into the membrane pores.
While wetting can be caused by the pressure in the feed channel exceeding the liquid entry pressure
(LEP), fouling induced wetting is the concern for real MD systems. Hydrophobicity of the membrane
material is the reason why the interior of the pores are not normally wetted. Scaling along the pores
with salt crystals growing into the pore tends to reduce the net hydrophobicity and non-wetting
character of the pores. Wetting caused by scaling is an important long-term performance issue for MD
since the maximum concentration of the salts is expected to occur close to the pore openings where
water evaporates, therefore the potential for precipitation is highest in this region.
Once wetting occurs, the MD process is affected in several ways. In several studies, water has
been observed to more easily penetrate adjacent pores [25, [59, 57, 26, 84]. Further crystallization can
also lead to accelerated wetting. Once the membrane is wetted, MD is no longer selective and hence
doesn’t achieve its goal of desalination or other types of separation. Figure 5 shows that scaling can
occur within the membrane pores following wetting. The presence of this layer of salt deposits renders
the top surface more hydrophilic, making it more prone to wetting [24, 111, 112]. In some cases, such
propensity to wetting was shown to affect only the top-most portion of the pores of a polypropylene
(PP) membrane, leaving the pores beneath un-wetted [112]. In other cases, wetting across the whole
membrane thickness was reported, which has been seen in PTFE and PVDF membranes, evidenced by
the presence of salt crystals at various depths of the membrane’s cross section [24, 102].
32
Fig. 5. Left: Cross sectional SEM image of salt deposits inside a PVDF membrane after the 4th week of
real seawater exposure [51]. Right: Cross sectional SEM image of deposits located inside a PP MD
membrane after rinsing with HCl solution [35].
Pore wetting may degrade the performance of the MD process either because it reduces the
interface for evaporation and therefore the production of vapor, or because, once a pore is wetted,
saline water may flow through and contaminate the distillate [24, 59, 57, 102, 112, 113].
Finally, one very interesting impact of scaling on MD membranes is the occurrence of “negative
flux,” reported by Guillen-Burrieza et al. [24] in one of their PVDF membrane tests. According to GuillenBurrieza et al. [24], one of their fouled PVDF membranes showed a flux from the permeate side to the
feed side, until a minimum trans-membrane temperature gradient of 10°C was reached, after which a
positive flux (from feed to permeate side) was obtained. They attributed this negative flux to the
osmotic pressure created by the localized high salinity at the membrane surface on the feed side due to
the deposited salt layer. Franken et al. [114] observed a 30% decay in DCMD flux over a period of one
month and postulated that this must be a result of membrane wetting and possibly “back flow.” Laganá
et al. [115] has also reported a similar phenomenon.
4.2 Permeate flow rate reduction
33
Relatively few studies report data on fouling and its effect on the MD process performance.
Different experiments report fouling differently, so a comparative metric is desirable. The average
percentage flux reduction is defined as the reduction in flux as a percentage of the initial flux. The rate
of change of the flux reduction can be used to examine the rate at which fouling lowers flux, as shown:
1234567!89 :
32=>=?=@A − 32=>@A 100
%
×
<=
32=>=?=@A
ℎ7
8
which may eventually level off at some system steady state. In a modeling context, flux decline is
associated with a decrease in the MD coefficient B. The MD coefficient, or B coefficient (kg m-2 Pa-1 s-1),
characterizes the permeability of a membrane under the MD process. The coefficient B is a function of
the membrane material properties (i.e. pore size, thickness, etc.) and the operating temperature [116].
The degree of permeate decline varies widely by experiment. In some processes, no decay
occurred after months of operation [3, 117], while in others flux decline was as high as 66% in less than
two days [59]. Permeate decline is dependent on the porosity of the scale that occurs and the thickness
of scale layer [26], thus it is largely dependent on the salt or type of scale, the local concentration, and
the scale’s solubility under the given operating conditions. Decline may be slow and gradual [118], or it
can occur rapidly from rapid crystal growth after exceeding critical levels of supersaturation [73]. As
discussed previously, uncharged fouling including biofouling and iron oxide fouling were found to be
relatively porous, while inorganic scale such as calcium carbonate were relatively nonporous.
Theory on fouling in membrane systems suggests different profiles for decline in flux over time
based on different models of fouling [119]. Srisurichan et al. [119] applied the cake model of scale
deposition to DCMD experiments using HA as a foulant with NaCl and CaCO3 in the feed and found that
it was capable of explaining the decline in flux. This paper illustrates that previous research on the use of
34
additional transport resistances to model a fouling layer can be adapted to MD systems, accounting for
the fact that the driving force in MD is a vapor pressure difference rather than a pressure difference.
Table 1 shows data gathered from publications that report fouling induced flux reduction. The
program WebPlotDigitizer [120] was used to analyze the graphs and estimate initial and final flux values
and the elapsed time between them. This data is used in subsequent analysis to draw conclusions about
the effect of system parameters on rate of fouling.
Table 1: Fouling induced flux changes reported in the literature.
Stu
dy
MD
configura
tion
Feed
Feed,
Permeate Initi
Inlet
al
Temperat Flu
ures [°C].
x
Permeate [L/
Pressure m2
(Pp) [kPa] day
]
for VMD
Ground Water
Ground Water + 0.1
mol/l HCl
60,
Pp = 1.5
40,
Pp = 3
60,
Pp = 1.5
60 ,
Pp = 1.5
5
Ground water GW
85, 20
6
Boiled GW
85, 20
7
Tap Water TW1
85, 20
8
Boiled TW1
85, 20
Tap water TW2
85, 20
Boiled TW2
Tap Water TW1 +
85, 20
85, 20
1
2
7 ppt NaCl solution
7 ppt NaCl solution
[12
1]
3
4
VMD
[72]
9
1
0
1
DCMDhollow
680
.2
391
.0
529
.7
529
.8
400
.0
395
.0
418
.0
412
.0
795
.0
720
.0
407
Fin
al
Flu
x
[L/
m2
day
]
597
.4
332
.2
376
.1
506
.9
350
.0
383
.0
247
.0
320
.0
705
.0
590
.0
404
Time of
experi
ment
[hours]
150.0
150.0
75.0
Additio
nal
Informa
tion
Tind =
100.3
hrs
Tind =
90.4 hrs
Tind =
40.9 hrs
70.0
23.0
Average
Fouling
Rate
D
%EFGHIJKLJMNJ
OP
0.08
0.10
0.39
0.06
vf =
0.4m/s
0.54
23.0
0.13
44.0
0.93
38.0
0.59
65.0
0.17
36.0
27.0
0.50
0.03
35
2
1
3
1
4
1
5
1
6
1
7
1
8
1
9
2
1
2
2
2
3
2
4
2
5
2
6
2
7
2
8
2
9
3
0
HCl pH = 4
DCMDhollow
[39]
[10
2]
[10
1]
[11
8]
[11
3]
DCMDhollow
DCMD
DCMD
Tap Water (CaCO3)
90, 20
Tap Water (CaCO3)
85, 20
Tap Water (CaCO3)
80, 20
Tap Water (CaCO3)
80, 20
Tap Water (CaCO3)
80, 20
Tap Water (CaCO3)
80, 20
Tap Water (CaCO3)
80, 20
Waste water
85, 20
Waste water boiled
85, 20
CaSO4 (batch mode)
Na2SiO3 (batch
mode)
60, 20
CaSO4 (batch mode)
60, 20
CaSO4 (batch mode)
50, 20
CaSO4 (batch mode)
40, 20
Seawater
MF treated
Seawater
60, 20
Raw Seawater
45, 20
60, 20
45, 20
DCMD
.0
729
.7
612
.0
548
.0
414
.4
476
.0
537
.7
599
.3
321
.0
364
.4
265
.3
265
.3
739
.2
496
.8
259
.2
570
.2
350
.0
230
.6
.0
653
.1
577
.0
534
.0
242
.5
416
.1
506
.4
479
.5
288
.0
348
.8
32.
9
172
.6
38.
4
100
.8
28.
8
344
.6
276
.1
185
.0
24.7
0.42
25.0
0.23
24.4
0.10
43.0
66.0
89.8
82.9
vf = 0.3
m/s
vf = 0.6
m/s
vf = 1.0
m/s
vf = 1.4
m/s
0.96
0.19
0.06
0.24
47.5
0.22
45.0
0.10
37.3
59.3
15.5
18.1
37.0
720.0
Tind =
28.6 hrs
Tind =
21.6 hrs
Tind =
4.4 hrs
Tind =
8.5 hrs
Tind =
30.1 hrs
vf = 0.14
m/s
2.35
0.59
6.12
4.40
2.40
0.05
168.4
0.13
167.3
0.12
4.3 Increased temperature and concentration polarization
Scaling
may
increase
temperature
and
concentration
polarization
by
creating
a
hydrodynamically stagnant or slow moving layer of water at the surface of the membrane [27, 111]. If
scaling impedes flow velocity, which occurs particularly in hollow fiber membranes, the slower velocity
will increase temperature polarization as the water residence time is lengthened [122]. The deposited
36
salt layer creates an additional thermal resistance, reducing the heat transfer coefficient from the feed
bulk to the evaporation and condensation interfaces [105], and in some cases possibly accelerating the
degradation of polymeric materials [111]. As explained in Section 2.3.1 Bacteria and Biofilms, a fouling
layer impedes convective heat transfer in the bulk. Although fouling layers have greater thermal
conductivities than the polymers composing the membrane, they do not increase temperature
polarization directly. Instead, they impart mass transfer resistance, altering the heat transfer associated
with evaporation and consequently increasing the temperature polarization [123].
The mass transfer coefficient close to the membrane surface is also often reduced due to the
presence of biofilms or other foulants. This leads to an increase in the concentration of dissolved ions
close to the membrane interface, reducing the local vapor pressure and thereby reducing flux, in
addition to increasing tendency to precipitate [124].
Recent studies by Goh et al. [123] found that hydrophilic biofouling on membrane surfaces that
reduce the average pore size of the evaporating surfaces can reduce the MD driving force by causing
vapor pressure reduction, as described by the Kelvin equation.
4.4 Membrane damage and chemical degradation
Many studies, especially at the lab-scale, have reported physical damage to the membrane as a
result of scaling [24, 35]. The damage to the MD membrane was observed to take several forms. These
include: a reduction in hydrophobicity of the membrane surface by altering its chemistry [24, 112], 119];
alteration of the membrane’s pore structure and pore size distribution [24]; reduction in the
membrane’s mechanical strength [24]; reduction in the membrane’s permeability via surface blockage
[24, 112]; and the formation of defects (e.g. cracks) within the membrane structure [24]. In most cases,
the damage was associated with a deterioration of the distillate quality (lower salt rejection) [22, 23, 35,
37
57, 77]. The most frequently reported membrane damage when scaling occurs during MD is the
formation of a scale layer on the membrane’s top surface in contact with the feed (Figure 6). This layer is
composed of insoluble salts, such as CaSO4, MgCO3 and CaCO3 [24], in addition to NaCl [24, 112].
Interestingly, membrane damage due to scaling can lead to either higher or lower flux than that
of an intact membrane. While lower flux can be attributed to pore blockage by scale deposits on the
surface [35, 112], higher flux is primarily due to pore wetting, usually accompanied by a lower salt
rejection [24, 77]. For example, Hsu et al. [113] reported severe fouling and flux decline, but without
permeate quality deterioration, when PTFE membranes were used to desalinate pre-treated subtropical seawater, with high NOM and biofouling potential. Ultrasonic cleaning was applied periodically
in this study, which restored most of the flux. However, the data suggest a small degree of irreversible
fouling, as evidenced by a slow reduction in the flux with time [113].
Guillen-Burrieza et al. [24] reported that the scale layer formed on the membrane surface during
MD operation reduced the gas permeability of the membrane [24]. However, MD experiments showed
apparently higher permeate fluxes for the fouled membranes which were attributed to heavy pore
wetting processes caused by the inorganic scaling and membrane damage. Post-deposition washing with
de-ionized water did very little to remedy that. Gryta et al. [112] reported similar findings for PP MD
membranes exposed to NaCl solution as feed. A similar behavior was also reported for fouled
membranes [105] when a biofouling protein layer was formed after concentrating saline wastewater.
Changes in membrane morphology upon fouling are the second observed damage in MD
membranes. Gryta et al. [112] noticed only a minor change in membrane porous structure in PP
membranes and concluded that the polymer material used for PP membranes production exhibited
good thermal stability [112]. This was not the case for the PVDF and PTFE membranes studied by
Guillen-Burrieza et al. [24], who conducted a study on the effect of cyclic wet-dry MD operation using
38
seawater. The latter reported a noticeable variation in membranes’ porosities and a shift in their pore
size distributions (PSD) upon fouling after two weeks of seawater exposure. The shift in PSD was more
pronounced in the PVDF membranes (which had a broader PSD to begin with) than in PTFE membranes.
They attributed this change to: i) buildup of a relatively thick (4-7 µm) salt deposit layer on the
membranes’ surface; and, ii) damage to the membranes, especially PTFE ones, in the form of cracked
fibrils (Fig. 6) and altered structure. The cracks in PTFE membranes, described as being similar to those
observed during shrinkage of dried clay, were attributed to the dry out periods. In PVDF membranes, on
the other hand, a buildup of salt crystals within the membrane was observed. Guillen-Burrieza and coworkers concluded that PVDF and PTFE behaved differently in their reaction to a fouling medium under
MD [24]. A series of AFM studies were also conducted by this group, which revealed that the two
membrane materials behaved differently in terms of their attraction forces to CaCO3 salt crystals, as well
as their surface roughness [24]. Collectively, this strongly suggests that the nature of membrane
material, in addition to its surface morphology, has an important role to play in resisting fouling in MD.
The mechanisms behind this role are yet to be explored.
39
Fig. 6. (a) and (b): Cracking and thick salt layer deposit on two different PTFE membranes under
intermittent drying conditions after 2 weeks of exposure to real seawater [24].
Using Mullen burst test, Guillen-Burrieza et al. [24] were able to demonstrate a third impact of
fouling on MD membranes: the loss of their mechanical strength. They showed a difference between the
Mullen burst pressure of PTFE and PVDF membranes before and after fouling. While PVDF membranes
remained unaffected after seawater exposure and maintained their integrity, fouled PTFE membranes
showed a decreasing trend in the pressure required to initiate the burst (even with the presence of the
PP backer material). This strength deterioration was attributed to the fibril damage and crack formation
seen in PTFE during the dry out processes.
Another reported form of membrane damage during MD operation is that which is due to
chemical degradation. Using Accurel PP membranes, Gryta et al. [112] reported significant
hydrophilization of the membrane surface, leading to increased rate of wetting. They found that the
operating conditions of their MD process, including elevated temperature and the presence of oxygen,
enabled the degradation of the PP polymer by forming hydrophilic groups (hydroxyl and carbonyl) on its
surface [112]. These surface groups reacted with the concentrated NaCl solutions and consequently
sodium carboxylate was formed. This was supported by FTIR analysis [112]. Attempting to regenerate
the wetted membranes by rinsing and drying was found ineffective, since the presence of those
hydrophilic groups caused a rapid membrane rewetting during the consecutive MD operation [112]. The
presence of salt, according to Gryta et al. [112], has also stiffened polymer molecules (i.e. PP),
preventing their chain disentanglement, and leading to chain scission. However, the quality of the
distillate in their study was found to have remained unaffected [112].
40
5 Fouling mitigation in MD
The main scaling prevention tools employed in MD are feed pretreatment and chemical cleaning
[1, 58]. Other fouling prevention methods attempted include increasing the feed flow rate, hydraulic
cleaning, reducing surface roughness, changing the hydrophobicity of the membrane, magnetic water
treatment, and changing surface charges on the membrane [16, 125]. The effects of the filtration and
antiscalants have been studied in MD, as well as less commonly used technologies like feed heating or
boiling, pH changes, flocculation, and magnetic water treatment [49, 59, 72, 82].
Pretreatment of the feed is standard practice in most desalination systems, and pretreatment
needs vary significantly by technology and feed water quality. Common pretreatment methods include
filtration, antiscalants, flocculation, and chlorination [107]. The market dominant desalination
technology, RO, has intense pretreatment demands to protect the membranes, and some reports
anticipate that MD will not need this level of pretreatment, as liquid water does not pass through the
membrane and no cake compaction takes place. [107]
5.1 Thermal water softening
Certain water conditions, such as groundwater with high hardness, may benefit from the
intentional breakdown of bicarbonate ions through high temperatures or boiling. This process is called
thermal softening, and helps reduce scale by causing CaCO3 and other salts to precipitate out in a
heating step, and also reducing CO2, which is linked to bicarbonate ions as explained previously. For an
MD system, M. Gryta boiled feed water for 15 min. and paper filtered it prior to undergoing hollow fiber
MD, which lowered the bicarbonate ion content 2 to 3 times, to a concentration of 1.5 mmol HCO3 -/L
[72]. While boiling is an expensive pretreatment option in terms of energy use, since MD needs a hot
feed stream, the boiled water is used in the MD setup after some salts precipitate out of the solution.
As a result, flux declined only 3% over 23 hours, instead of 12% without boiling. However, for tap water
41
with a concentration of 0.4-0.95 mmol HCO3-/L, CaCO3 fouling was worse than in the untreated case
[72]. A comparison of the rate of fouling is shown in Fig. 7. Although not addressed in the paper, boiling
likely reduced the concentration of dissolved CO2, which in turn made calcium carbonate more likely to
precipitate, and that may have been the cause of this non-intuitive result that contradicts other typical
results. Boiling may therefore only be advisable with certain feed water solutions for certain durations.
1.00
% reduction in flux/hr
0.80
0.60
0.40
0.20
0.00
Ground water GW
Tap Water TW1
Tap water TW2
Before
0.54
0.93
0.17
After
0.13
0.59
0.50
Fig. 7. Literature data. The effect of boiling as pretreatment on the fouling rate (represented as the % of
flux reduction per hour) for different source waters (see Table 1 for references).
5.2 Micro/nano filtration pre-treatment
Microfiltration and nanofiltration (NF) is sometimes used before the MD process to remove
particulate material and large molecules in the feed [87, 126]. Filtration may also be used to remove
scaling the bulk, such as in the study by Kesieme et al where a cartridge filter captured calcium scale,
allowing high recovery of groundwater RO concentrate with MD [127]. While ultrafiltration is good for
removing suspended solids and colloids, NF can be an effective pretreatment for rejecting organic
matter and removing hardness from feed water. As commonly practiced in RO, some researchers have
42
also suggested using chemical coagulation followed by sand filtration or microfiltration to decrease
fouling potential in MD [72]. Lawson and Lloyd [3] note that “several investigators … reported a
reduction in the degree of flux decay for pre-filtered (≈1 µm) process liquids” [3, 117, 128, 129].
Alkalaibi and Lior [130] have observed that following pretreatment with nanofiltration and acid addition
to a pH of 5, the microorganism count in the water was close to zero even after 1400 hours.
Flocculation has been examined in conjunction with microfiltration for wastewater MD
treatment, which found microfiltration to be very effective in reducing suspended solids, with relatively
mild to weak improvement from FeCl3 coagulation/flocculation [82].
5.3 Use of antiscalants
Antiscalants are a common tool to prevent inorganic scaling, and are potent for carbonate
scales, as well as phosphate, sulfate, and fluoride, disperse colloids, and metal oxides [39, 131, 132].
Antiscalants act through a variety of mechanisms, including delaying nucleation, reducing the
precipitation rate, distorting crystal structure, and altering CO2 concentration [133]. They are generally
the most common technique for scale control because of low costs, and usually require dosing of less
than 10 ppm [133]. However, antiscalants molecules, typically organic, often reduce the surface tension
of the water, which can promote membrane wetting [134].
A notable antiscalant study was performed for hollow fiber MD using a polyphosphate-based
antiscalant designed for RO, and compared with laboratory grade sodium polyphosphate [28]. The
particular antiscalant works by sequestering calcium, thus inhibiting precipitation. The name and
composition of the commercial Polyphosphate based anti-scalant were not given. With the antiscalant,
the formation of CaCO3 crystals was virtually eliminated, but a thin amorphous non-porous layer
deposited on the MD membrane. Surprisingly, the flux decline was worse when the antiscalant was
43
present. The higher the antiscalant concentration, the lower the permeate flux. However, the associated
scale was mostly on the membrane surface, so periodic HCl cleaning was effective. Because of
orthophosphate deposit risk from the breakdown of this antiscalant, residence times in MD were
recommended not to exceed one hour [28]. Gryta concluded that regular HCl cleaning with an
antiscalant system could make it useful in MD [28]. A different study by He et al. [135] found more
positive results using polyacrylic acid antiscalants in MD, which were particularly effective in reducing
calcium sulfate scale. Organo-phosphorus antiscalant compounds were also very effective in reducing
calcium carbonate scale and mildly effective in reducing calcium sulfate scale [135]. An antiscalant
blend of carboxylic and phosphoric acids was moderately effective in reducing both calcium carbonate
and calcium sulfate [135]. Surface tension of the proposed antiscalant solutions (concentrations varying
from 0.6 to 70 mg/L,) was very close to that of tap water (71.5 mN/m against 71.8 mN/m for tap water),
and no wetting phenomena were detected under the experimental conditions.
5.4 pH Control of the MD feed
The pH control of the feel has been a common method to reduce or eliminate MD scaling. In
almost all cases, this means acidifying the feed, as alkaline salts, the main component of scale, become
drastically more soluble at acidic pH. However, according to Karakulski’s findings, acidification of the
feed failed to prevent silica scale [59, 110]. In one DCMD hollow fiber study with solutions of CaCO3 at SI
of 49 and CaSO4 at SI 1.12, acidification to a pH of 4.1 fully prevented scale under conditions that
experienced rapid scaling of MD membranes at neutral pH [50, 59, 110]. Numerous researchers found
that bringing the feed to a pH of 4 with acid addition was sufficient to virtually eliminate calcium
carbonate scale even at extreme SI indices [50, 72, 110]. Figure 8 shows the relative reduction in fouling
rate (calculated as the rate of % flux decrease per hour, see Section 4.2) achieved by controlling the pH
44
of the feed solution. However, adding solution to modify the pH can quickly become prohibitively
expensive, depending on the pH [136].
1.00
% reduction in flux/hr
0.80
0.60
0.40
0.20
0.00
Gryta2010 pH=4
Pangarkar2011 pH=1
Before
0.93
0.39
After
0.03
0.06
Fig. 8. Literature data. The effect of adding HCl to the feed solution and reducing pH as pretreatment on
the fouling rate (represented as the % of flux reduction per hour) [72, 83].
5.5 Magnetic water treatment
Magnetic water treatment (MWT) is a technology developed for scale reduction in water
treatment [49, 137, 138] and power plant heat exchangers [49]. The magnetic field slows nucleation
while increasing crystal growth rate [112, 113, 139, 140], and it can alter the precipitate morphology and
properties. Gryta performed a study with a commercially available MWT device called Magnetizer RWES on hollow fiber MD with a tap water feed [49]. The experiment ran with a 0.1 T magnetic field, feed
temperature of 85°C, and alkalinities of concentrations of 2.72 and 3.61 mmol/dm-3 HCO3-. The magnetic
field caused significant changes in the morphology of crystal deposits on the membrane, causing crystals
to be larger, but the deposits to be more porous. MWT caused the deposit layer thickness to be 10-25%
smaller, and significantly mitigated flux decline. It was expected from previous studies that MWT would
45
shift CaCO3 crystal type towards vaterite and aragonite forms, but under the temperature and slow
laminar conditions, predominately calcite deposits occurred [49]. No effect on membrane wettability
was found.
5.6 Tailoring MD membrane properties
Increased hydrophobicity of MD membranes has shown to have a dramatic effect in reducing
scale formation. Hydrophobicity differences between materials may also account for different
propensities to foul. For example, according to Gryta, polypropylene membranes do not have an optimal
hydrophobicity, and undergo wettability readily, often after a few days in his experiments [111].
Superhydrophobic coatings provide an additional hydrophobic layer, which acts as a buffer layer
from scale
[141] by reducing the surface nucleation and the particulate attachment.
Superhydrophobicity can reduce membrane wetting and improve membrane recovery from acid
cleaning [125, 142]. One method of creating superhydrophobic MD membranes is depositing TiO2
nanoparticles using LTH (low temperature hydrothermal synthesis). The superhydrophobic membranes
generally have fluorosilicone coatings, and may achieve contact angles of 166° [125]. Coating the
membranes with fluorosilicone for hollow fiber MD was found to create very long induction times of 194
minutes for gypsum [135], compared to no induction period otherwise. Other sources found significant
CaCO3 scale reduction [27, 50] and CaSO4 scale [50] using fluorosilicone coatings [143]. These coatings
can also reduce pore sizes, increasing resistance to scale [50].
As mentioned previously, superhydrophobicity also has shown to prevent microorganism
fouling. However, more hydrophobic membranes are known to preferentially absorb HA [25]. Studies by
Meng et al. found that superhydrophobic PVDF membranes fouled similarly to humic acid as ordinary
PVDF, despite significantly reduced inorganic scaling [95]. According to Meng et al., humic acid fouling
46
on the superhydrophobic membrane occurred via an adsorption-desorption mechanism. A commonly
used method to prevent biofouling is based on the hydrophilization of the membranes. Using sodium
alginate hydrogel as coating may reduce the adsorption of organic compounds such as citrus oil on PTFE
membranes [104]. Additionally, UV-induced grafting of zwitterionic polymers on PP membranes has
been used to prevent protein fouling with very good results [144] as well as interfacial surface
crosslinking of PP membranes [145] without compromising the rejection factor.
Increased surface porosity of membranes seems to be a factor in increased tendency to scale as
well. Gryta performed a study of the effect of porosity on membrane performance using polypropylene
capillary membranes and tap water. He found that the presence of larger pores on the surface allowed
for the deposition of CaCO3 crystals in the membrane interior, causing wettability, especially wetting
during HCl cleaning [111]. The surface with higher porosity was found to have significantly increased
wettability. However, no difference in flux decline was observed between the highly porous and nonporous surfaces of otherwise identical membranes, indicating that high porosity may not increase
clogging or membrane blocking [111]. Gryta found that membranes with a low porosity coating 1 µm
thick has significantly less tendency to wet, while exhibiting similar average properties and permeate
production to the uncoated membrane [111]. Gryta inferred that crystal growth inside the membranes
can be restricted by pore diameters much smaller than the crystal size. Such low porosity coatings may
therefore be recommended to avoid wettability with only minor reduction in permeate production.
Finally, some studies suggest that membrane material type may have a significant impact on
scaling. Curcio et al. found that in the presence of polypropylene hollow fiber membranes, the induction
time for CaCO3 was 18 seconds, compared with over 80 seconds for no membrane present [25, 53, 54].
Tung-Weng et al. [71] note that PVDF membranes wet more than PTFE. Non-polymer membranes such
as glass membranes may have superior thermal and chemical robustness [142].
47
5.7 MD module design and operation
The conditions within the MD module may also significantly affect fouling. Temperature and
concentration polarization may cause scaling preferentially at the membrane. Residence time in the
module may have an effect as well, as stagnation areas can promote scalant precipitation from the bulk.
Scaling is caused by conditions on the saline feed side rather than pure permeate, so system
configuration (e.g., AGMD, DCMD, VCMD, or SGMD) plays a role only because they have different flux
magnitudes [146]. Technologies with more conduction losses and with higher permeate flux, such as
DCMD [116, 147] or VMD respectively, may be expected to have somewhat worse scaling issues from
temperature and concentration polarization effects, as explained below.
However, based on a
numerical model developed by the authors elsewhere [148], the evolution of the Saturation Index (SI) as
a methodology to predict extent of scaling, has been calculated and detailed. Understanding the theory
associated with inorganic salt precipitation can help design safe operating conditions where the extent
of precipitation can be controlled.
5.7.1 Temperature polarization
Temperature polarization is one of the most important secondary phenomena that affect the
MD process. Water that evaporates into the pores of the MD membrane removes the corresponding
latent heat of evaporation from the liquid feed stream. This cools down the fluid close to the membrane
and results in a thermal boundary layer. Since the vapor pressure of water rises exponentially with
temperature, any reduction in temperature of the water at the membrane interface leads to a
significant drop in MD driving force. As a result, care is often taken in MD design to reduce and limit this
effect as much as possible. At a given flux, the most important factor that determines the extent of
temperature polarization is the effective heat transfer coefficient in the feed stream. At a higher heat
transfer coefficient, a smaller ∆T is required across the thermal boundary layer to transfer the same
48
amount of heat of vaporization. The most common design strategies are to increase the flow rate of
feed, operate in the turbulent regime or have turbulence promoters in the form of a spacer to increase
the heat transfer coefficient on the feed side.
The simplest model for temperature polarization is given by Eq. 1, where h is the heat transfer
coefficient in the feed channel and QR? is the total heat loss across the membrane including sensible
heat transfer and latent heat of evaporation of the vapor [12]:
, − , =
T UVW
X
(1)
In general, temperature polarization reduces vapor flux, but it may also reduce scaling tendency
since many critical salts’ solubilities vary inversely with temperature. Conditions giving rise to a large
temperature polarization may also create a large concentration polarization, which has the opposing
effect of raising concentration near the membrane and potentially promoting scale formation.
5.7.2 Concentration polarization
The MD process allows only water vapor to pass through while retaining non-volatile salts,
causing an increase in salt concentration near the membrane which is referred to as concentration
polarization. A simple model of the concentration polarization process is the film model [17] described
by Eq. 2:
YZ
Y[
^
= 9\ ]_`a
(2)
where J is the water flux through the membrane, k is the mass transfer coefficient for the salt in
solution, and xm and xc are the molar concentrations at the membrane interface and the bulk of the fluid
streams, respectively.
49
Concentration polarization affects the MD driving force through its effect on vapor pressure, but
this influence is minor relative to the effect of temperature polarization on vapor pressure. The more
important effect of concentration polarization is that the elevated concentration may promote scaling
formation on the membrane. As can be seen from Eq. 2, the effect rises with an increase in flux and falls
with an increase in the mass transfer coefficient. It should be noted that temperature and concentration
polarization are coupled phenomena since both depend on and influence the water vapor flux through
the membrane.
5.7.3 Saturation index, numerical modeling
Thermodynamically, membrane fouling is expected to be a function of the salt supersaturation
at the membrane surface. SI is a commonly used measure of supersaturation [149]:
hijklkjmnopqrij
v
stu
bc = logg
= (, )
(3)
(4)
According to these equations, temperature and concentration have an effect on the
thermodynamic tendency for precipitation. These two parameters also influence the permeate flux and
are in turn influenced by the flux and heat transfer rates (Eq. 1, 2). Engineering Equation Solver (EES)
was used to numerically solve the implicit set of equations that connect these parameters along with
the mass and energy balance equations, in order to predict the value of the SI on the membrane
interface and feed bulk. The modeling methodology is described in detail in [12, 148]. In the following
sections, results from the model are used to better understand the trends reported in literature. The
numerical modeling provides a quantitative understanding of the competing effects and the reason for
the overall trend observed. Data on changes in the extent of membrane fouling with modifications in
feed flow velocity and feed inlet temperature are available in literature. In the following sections, data
50
from reference [39] is used to illustrate the effect of feed flow rate and feed inlet temperature on the
extent of CaCO3 fouling induced flux decline. Modeling results for an AGMD setup on the effect of
system variables on SI at the membrane interface are also presented.
For the numerical model the following assumptions have been made: an inlet salinity of 13
mg/kg is assumed which is approximately equal to the solubility of CaCO3 at room temperature;
diffusivity of CaCO3 is assumed to be 1.469 × 10-9 m2/s and the dependence of solubility product of
aragonite on temperature is modeled as in [39] (T is temperature in K):
= −171.9773 − 0.077993 +
"yz'."y'
+
{
71.595 log (5)
It is important to note that only the trends in SI variation with changing system parameters are
being compared with the reported experimental fouling data. The actual value of SI in the experiment is
likely to be different due to differences in concentration of the feed water, presence of other ions in the
feed and differences in geometry and MD configuration.
5.7.4 Effect of temperature
As discussed previously, temperature has a dominating effect on scale formation due to salt
solubility variation with temperature. Similarly, temperature polarization may have a significant effect
on scaling and fouling since the solubilities of common foulants are highly temperature dependent, and
since biological fouling as discussed previously can be limited by temperature. Also, the effect varies
greatly depending on the salt. Reduced membrane temperature due to temperature polarization might
prevent precipitation of CaCO3 and CaSO4 which are less soluble at higher temperatures, but may make
non-alkaline salts scale more readily. Higher temperatures as stated previously may limit bacteria
growth, so high degrees of temperature polarization may enable biofouling. Figure 9 shows data on the
effect of temperature on fouling rate in an experiment where the major scalant was CaCO3 [39].
51
% reduction in flux/hr
0.5
0.4
0.3
0.2
0.1
0
75
80
85
90
Feed Temperature (oC)
95
Fig. 9. Literature data by Gryta. The effect of the feed temperature on the fouling rate (represented as
the % of flux reduction per hour) for Lake water with bicarbonates for hollow fiber DCMD [39].
52
Due to the inverse solubility of the salt and increased flux contributing to higher concentration
polarization in the feed stream, there is a significant increase in fouling rate with change in feed
temperature. The rate of fouling induced flux reduction is four times higher at 90oC than at 80oC.
To numerically explain this behaviour, the previously described EES model was used to estimate
the effect of increasing feed temperature on the SI at the membrane interface and feed bulk. The
simulations were carried out for a system with a feed flow rate of 0.1575 kg/s in a feed channel with a
depth of 4 mm and a width of 12 cm. Figure 10 shows the predicted variation in SI at the membrane
interface and in the bulk of the feed stream.
Fig. 10. Simulation results at fixed,=> = 0.1575kg/s. Saturation Index of CaCO3 in the feed stream at
the membrane interface (SIf,m) and in the bulk (SIf,b) as a function of feed inlet temperature. Increase in
SIf,m correlates with higher scaling observed. Sif,m increases faster than SIf,b illustrating the effect of higher
concentration polarization.
53
With an increase in feed inlet temperature, the SI of the bulk fluid increases associated with a
reduction in the Ksp for CaCO3 (Aragonite). At the membrane surface, the increase in SI is higher, since
there is a larger water flux and an associated increase in membrane concentration of ions compared to
the bulk. The increase in predicted SI at the membrane surface correlates well with the increased fouling
rate at high temperatures as illustrated in Fig. 9.
5.7.5 Polarization abatement: feed flow rate and bubbling
Feed flow velocity has a direct influence on the heat and mass transfer coefficients in the feed
channel. With an increase in feed velocity, both heat and mass transfer coefficients increase, and
transition from laminar to turbulent flow may cause this change to be discontinuous. As a result, with an
increase in flow rate, flux increases as a result of reduced temperature and concentration polarization.
For inverse solubility salts such as CaSO4, Ksp decreases with an increase in temperature (Fig. 1).
Since the heat transfer coefficient (h) increases with flow rate, the temperature at the membrane
surface increases. At the same time, an increase in mass transfer coefficient will lead to a decrease in
salt ion concentration at the membrane surface. Since both the numerator (activity product) and the
denominator (Ksp) of Eq. 4 decrease, the relative change in SI is determined by the rate at which the two
quantities change.
Numerical modeling was again used to understand the overall effect of increasing the feed
velocity on the SI at membrane interface. The parameters used for the study were the same as
mentioned earlier, with the feed inlet temperature fixed at 80°C. Results are shown in figures 11 and 12.
54
Fig. 11. Simulation results at fixed,=> = 80°C. Effect of feed mass flow rate (m
f,in)
and Reynolds
number (Ref) on temperature (Tf,m) and CaCO3 concentration (Xf,m) at the membrane interface. Increase
in temperature would increase tendency for precipitation of CaCO3, whereas decrease in concentration
would reduce tendency for scaling.
55
Fig. 12. Simulation results at constant,=> = 80°C. Overall effect of the feed mass flow rate (m f,in) and
Reynolds number (Ref) on the SI at the membrane interface (SI
f,m).
Decrease in SI correlates with
experimentally observed decreased rate of fouling (Fig. 13).
Salt ion concentration decreases and temperature at membrane surface increases with increase
in feed flow rate (Fig. 11). The overall effect of these two opposing effects was a decrease in SI at the
membrane surface with increased feed flow rate, as shown in Fig. 12, since the reduced concentration
polarization leads to a larger decrease in ion activity product as compared to the decrease in Ksp
associated with lower temperature polarization. This result is consistent with data from Gryta [39]
(Table 1, Fig. 13) which shows that an increase in feed velocity from 0.3 m/s to 1 m/s leads to a
reduction in the rate of flux decline.
% reduction in initial flux
1.2
1
0.8
0.6
0.4
0.2
0
0
0.5
1
1.5
Feed Velocity (m/s)
Fig. 13. Literature data. The effect of the feed flow velocity on the fouling rate at constant feed inlet
temperature of 80 °C (represented as the % of flux reduction per hour). Decrease in fouling rate with
increase in feed velocity is shown [39].
56
The marginal increase in fouling rate between 1 m/s and 1.5 m/s (feed flow velocity), is
inconsistent with the theory presented above and could be associated with transition to turbulence that
is expected to happen at these velocities in the experiment.
In addition to the thermodynamic effects discussed above, at higher flow rates, the shearing
action of the water in removing deposited precipitates is higher. This can help reduce the overall
membrane scale deposition and fouling rate. Gryta [39] also reported that the size of salt crystals
observed was much larger at the lower flow rates (0.31 m/s) as compared to higher flow rates (0.62
m/s), which can be explained by considering the kinetics of the fouling process as explained above.
Other methods to reduce concentration and temperature polarization have also been explored.
Recently, Chen et al. [150] have demonstrated the effectiveness of gas bubbling through the feed as a
means of promoting mixing and thereby improving flux and reducing scale deposition. Traditional
methods to promote better mixing include increased flow rate and the use of spacers to induce
turbulence in the feed channel and to improve mixing. Chen et al. compare three different
configurations of hollow fiber membrane MD with feed outside the fibers: no-spacer, spacer and air
bubbling without spacer. Figure 14 shows the type of spacer used in that study.
Fig. 14. Example of mesh spacers used in hollow fiber MD experiments [150].
57
The flux is lower for the no spacer case as compared to the other two. With time (about 7 hours
of experiment), when the bulk NaCl concentration by weight is about 23%, a significant fouling layer and
associated decline in flux rate is observed in all three cases. While the air bubbling case showed less
scale formation, the spacer actually leads to more scaling than the no-spacer case, since it enables local
trapping of NaCl close to the membrane. Air bubbling achieves the goal of mixing the feed stream and
improving performance, while at the same time retards local scale formation. Amongst the air bubbling
cases, performance improvement is higher for cases where significant temperature or concentration
polarization is expected (at higher feed water temperature and laminar flow conditions). Air bubbling
will also increase the dissolved gas concentrations in the feed which has potential impact on gas content
in the membrane pores, permeate-side mass transfer resistance (for AGMD or SGMD), and on the
formation of salts depending on the gas chemistry (e.g., CO2 content) [88].
5.7.6 Types of module
Several MD module configurations have been designed and studied for fouling, with significant
differences. Although direct comparisons are lacking, evidence from similar feed solutions indicates
dramatic differences in fouling risk. The types of MD modules include: hollow fiber membranes which
consist of small tubular capillaries; flat plate modules; tubular modules which consist of concentric
cylinders that resemble flat plate membranes; and spiral wound membrane modules, which consist of a
membrane and spacer wrapped around a collection tube [1]. The largest differences can be found
between hollow fiber membranes, which have been studied for MD fouling relatively extensively, and all
the rest. Hollow fiber membranes are known to have higher fouling potential [50], but the present
review has found that their fouling potential is often extremely high, occurring under many conditions
for which other units have not been known to foul, and they have even experienced fouling with bulk
concentrations well below saturation conditions, which is highly unusual [3, 151]. Crucially, this high
degree of fouling has generally been found in the experimentally more common setup in which feed is
58
inside the capillary tubes [28]; in other experiments with distillate inside the capillary tubes, fouling was
in some cases regarded as unlikely [143], although it still can occur, as shown in Fig. 15.
Fig. 15. SEM images of PVDF Hollow Fiber DCMD fouled modules in cross-section with NaCl taken from
18 wt.% to saturation after 7 hours of operation with feed external to the capillary tubes [150].
In the case of hollow fiber membranes, the fine fibers cause slow internal laminar flow and
often extreme concentration polarization. They are also known to clog readily by colloidal particles in
the feed. Additionally, hollow fiber MD modules are difficult to clean and maintain [1], and if the feed
penetrates through the membrane pores, the whole module should be replaced.
Flat plate modules have been examined thoroughly as well in this review, and while tubular and
spiral wound modules have been investigated less, they are essentially flat plate systems with mild
curvature, and should not behave much differently. Flat plate modules are relatively easy to clean [1].
However, since hollow fiber and spiral wound modules can hold greater membrane areas they are
regarded as more cost effective MD modules. Therefore, cleaning protocols should be adapted to the
most commercially successful module option.
The various types of MD operation, including DCMD, AGMD, SGMD, and VCMD are not thought
to significantly affect MD scaling, but their differences on permeate flux and heat conduction alters
59
concentration and temperature polarization effects. These configurations differ on the permeate side,
not the saline feed side, so effects on fouling are secondary feed design [1].
5.7.7 Membrane cleaning
Acid cleaning is one of the most common methods used for fouled membrane reclamation.
Several studies have used different types of acids (both strong and weak) to remove scale components
from the membrane surface. HCl is by far the most commonly used acid in MD experiments for
membrane cleaning. It is particularly effective in removing basic salts such as CaCO3 by dissolving them
[39, 57]. Acid cleaning is carried out as a batch process with the feed water being replaced by an acidic
solution. During this period, the permeate should be discarded, as noted by Gryta [26, 152] since HCl is
volatile and can be carried into the permeate side along with water vapor. This was observed in a DCMD
experiment with hollow fiber membranes [26].
Gryta [28, 114] reported the efficacy of HCl rinsing in restoring flux back to original value in the
case of membranes with CaCO3 scale layer. However, after repeating the washing procedure with 2-5%
HCl, Gryta found that maximum restored flux declined [111]. Yang et al. [56] were able to restore flux
completely by cleaning with a 5% HCl solution following the use of untreated ‘tap water’ in their
experiments. Curcio et al. [25] used a two-step cleaning strategy using a citric acid solution followed by a
NaOH solution, allowing each of them to act for 20 minutes. They reported complete recovery of both
flux and hydrophobicity in an MD experiment using synthetic seawater. Some authors [29] have also
reported cleaning using bases such as 0.1 M NaOH for 20 minutes followed by pure water run for 2
hours. While pure water achieved 87.5% flux recovery, the NaOH solution was able to restore the flux to
the initial value. Gryta et al. found that pretreatment using Ca(OH)2 helped mitigate fouling from
silicates and sulfates [102].
60
Gryta [26] used HCl also to clean a membrane with iron oxide scale layer. It was found that 18%
HCl solution cleaning brought the flux close to the initial value and did better than 5% or 36% HCl. The
experiment showed that iron oxides were very difficult to remove from the MD hollow fiber membrane,
requiring high levels of acidification (36 wt. %) for 1 hour cleaning duration to obtain full removal, which
damaged the membrane [26]. Complete removal of the iron oxide scale did not eliminate all the
negative effects of fouling, which included decreased membrane flux after cleaning and increased
membrane wettability. Moreover, drying was needed to actually recover the flux and restore the
distillate quality because both were hindered (i.e., flux reduction from 800 to 650 dm3/m2-day and
distillate electrical conductivity of 20 µS/cm) after the acid cleaning. Therefore, Gryta recommends only
partial removal of iron oxide scale.
Table 2 shows data gathered from literature on the effectiveness of various cleaning protocols
for restoring permeate flux.
Table 2. Common cleaning methods used in MD and reported flux recovery.
S
No
1
2
3
3
4
Ref
Membrane
Material
PTFE
[121]
PP-hollow
fiber
[72]
5
[25]
6
Solution type
7 g/L NaCl
7 g/L NaCl
Ground Water
Ground Water (CaCO3)
Tap water (CaCO3)
Cleaning method
Water wash
Water wash
Water wash
3 wt.% HCl
3 wt.% HCl
Boiled TW (CaCO3)
3 wt.% HCl
20 min citric acid followed
by 20 min NaOH two stage
cleaning
Piezoelectric transducer at
35kHz used to induce
cavitation and cleaning of
membrane. Initially flux
restored.
PP-hollow
fiber
Seawater (HA, CaCO3)
PTFE
Seawater
[113]
7
Recovery
%
98.48
97.48
94.43
98.75
98.56
98.79
≈100
91.20
61
[27]
8
[39]
9
PP-hollow
fiber
PP-hollow
fiber
Microbial Biofilm
Ground Water
(primarily CaCO3,
CaSO4)
PP-hollow
fiber
Tap Water (CaCO3)
NaOH at pH=12, 40C
distilled water, 70%
ethanol for disinfection,
distilled water
2-5 wt.%HCl
[58]
10
5 wt.% HCl
≈100
≈100
Recovery
reduces
with
number
of
cleaning
cycles:
Cycle 1 99.9
Cycle 6 96.5
The effectiveness of acid washing is known to vary dramatically by the type of fouling. Alkaline
solutions that are more soluble at lower pH have been found to be very effectively removed by acid
cleaning, including the most common scale, CaCO3. Other crystalline scale, like iron oxide, has required
very strong acid to remove, [26] as described above. Silica scale has proven to be similarly difficult to
remove. Acid cleaning has caused mild but incomplete removal in cleaning organic matter on MD
membranes [30].
Sometimes cleaning has been achieved by simply running de-ionized (DI) water through the
system. DI water readily absorbs salts, and is even known to leach salts from surroundings [121]. Mericq
et al. [109] for example, completely restored membrane flux using RO permeate water to remove and
re-dissolve scale deposits from synthetic seawater in their VMD system.
In terms of other cleaning methods, a common and simple approach is reversing the flow
direction, which successfully restored flux from biofouling in AGMD Memstill experiments with pond
water performed by Meindersma et al. [87].
62
6 Trends in scaling in MD
Future developments in MD technology and industrial applications will create new issues and
areas needing investigation for MD fouling. Multi-stage designs of MD promise much better efficiency,
with some theoretical studies claiming lower energy consumption that the existing state-of-the-art
thermal technologies, such as MSF and MED [7]. Fouling in staged systems and for energy recovery
devices that recirculate fresh feed into later stages needs further research [26, 84]. As MD is developed
industrially, field experience and optimization of pretreatment requirements will be developed.
Additionally, as membranes designed specifically for MD are created, studies fouling with those
materials, porosities, B values, and other properties will be critical. MD use will continue to grow as well
in non-desalination areas such as the food, chemical, and dye industries [153], which were not
considered in this review. One final area of for further study is monitoring of scale development in
plants operations and long-term experiences.
7 Conclusion
Scaling and fouling in MD are found to be pervasive, but design and mitigation methods have
proven effective at making MD technology resistant to scaling and fouling. Four principal types of
fouling and membrane damage have been found in MD: inorganic salt scaling or precipitation fouling,
biofouling, particulate fouling, and chemical degradation. Inorganic scaling risk, the primary focus of
academic studies, varies greatly with the salts present. Alkaline salts such as CaCO3, the most common
scale by far, have proven to be readily prevented by decreasing feed pH or removed through acidic
cleaning, while other scale has proven more tenacious and must be generally be limited by avoiding
supersaturation. Biofouling has also been observed in MD, but can be largely mitigated through control
of operating conditions. Particulate fouling in MD has proven difficult to remove, but it can largely be
prevented by ultra- or microfiltration. Chemical degradation and damage to the membrane has proven
63
to be a concern as well, but can be mitigated by selecting operating conditions that avoid fouling,
extreme pH, and certain salts. The choice of membrane material and properties can also help to avoid
chemical degradation; PTFE and PVDF membranes exhibited different fouling characteristics, such as in
wetting, internal crystal growth, cracking, and mechanical strength [24]. A study on glass membranes
claimed glass membranes had superior thermal and chemical robustness than polymer membranes
[142]. However, there is a general lack of information about the effect of the polymer type on the
prevention of fouling/scaling in MD processes.
Fouling tendency had been perceived to be highly variable and perhaps unpredictable, but some
consistent patterns are seen. Studies with extreme susceptibility to fouling have almost exclusively been
performed with hollow fiber capillary membranes with the feed internal to the capillary tubes. These
modules have fouled within hours to days in unsaturated conditions that would not cause fouling in
other modules. Likewise, the studies showing high resistance to fouling tended to have highly
hydrophobic membranes or coatings, and include hollow fiber studies with permeate in the capillaries
[143]. Numerous studies have found substantial reduction in scale from superhydrophobic fluorosilicone
coatings, and while the individual papers may question how large a role the coating played in the often
complete lack of scale [50], the literature overall proves consistently that these coatings have a dramatic
effect. More hydrophobic membranes show higher resistance to wetting and associated internal crystal
precipitation. Since internal fouling often leads to further wetting, this too is reduced. Studies have
shown that increased hydrophobicity increases LEP and, in experiments, is associated with reduced
wetting, fewer fouling deposits, and purer condensate [125].
Micro, nano, or ultrafiltration has proven effective in stopping particulate scale. Modifying pH in
the feed or with cleaning may prevent or remove certain types of fouling very effectively as well.
Keeping feed temperature above 60 °C has proven very effective in mitigating biofouling, with some
64
exceptions. Removal of oxygen via deaeration may be expected to reduce biofouling and the authors
recommend further investigation into the effectiveness of this technique. Rinsing with a basic solution
such as with NaOH may resolve fouling for some substances, including HA. Mildly effective fouling
prevention methods include boiling for removal of carbonate, ultrasonic cleaning, magnetic water
treatment, flocculation, covering the membrane surface with a less porous smaller pore size layer, and
for HA, oscillating the feed temperature. Antiscalant effectiveness studies in MD have been inconclusive;
both strong reduction in scaling and actual decreases in permeate flux have been reported.
System design characteristics also influence fouling. Concentration polarization, closely related
to feed Reynolds number and rate of permeate production, is critical in causing fouling, and can be
mitigated by increasing the feed flow rate, or mixing technologies such as bubbling [150]. Temperature
also remains important, as the most likely foulants have inverse solubility with temperature. Simple
computational models were applied by the present authors to illustrate the effect of coupled heat and
mass transfer on scaling. A correlation between theoretical prediction of higher SI at the membrane
interface and increased rate of fouling induced flux decline was observed. Choice of safe operating
conditions should therefore consider temperature and concentration polarization effects and solubility
characteristics of the salts. Finally, stagnation zones or high residence times in the module may
contribute to fouling as well.
Acknowledgements
This work was funded by the Cooperative Agreement between the Masdar Institute of Science
and Technology, Abu Dhabi, UAE and the Massachusetts Institute of Technology (MIT), Cambridge, MA,
USA, Reference No. 02/MI/MI/CP/11/07633/GEN/G/00.
65
References
[1]
A. Alkhudhiri, N. Darwish, and N. Hilal, “Membrane distillation: A comprehensive
review,” Desalination, vol. 287, pp. 2–18, 2012.
[2]
M. El-Bourawi, Z. Ding, R. Ma, and M. Khayet, “A framework for better understanding
membrane distillation separation process,” Journal of Membrane Science, vol. 285, no. 1-2,
pp. 4–29, 2006.
[3]
K. Lawson and D. Lloyd, “Membrane distillation,” Journal of Membrane Science, vol. 124,
no. 1, pp. 1–25, 1997.
[4]
C. Cabassud and D. Wirth, “Membrane distillation for water desalination: How to choose
an appropriate membrane?,” Desalination, vol. 157, no. 1-3, pp. 307–314, 2003.
[5]
J. Gilron, L. Song, and K. K. Sirkar, “Design for cascade of crossflow direct contact
membrane distillation,” Industrial & Engineering Chemistry Research, vol. 46, no. 8, pp. 2324–
2334, 2007.
[6]
H. Lee, F. He, L. Song, J. Gilron, and K. K. Sirkar, “Desalination with a cascade of cross-
flow hollow fiber membrane distillation devices integrated with a heat exchanger,” AlChE
Journal, vol. 57, no. 7, pp. 1780–1795, 2011.
[7]
F. He, J. Gilron, and K. K. Sirkar, “High water recovery in direct contact membrane
distillation using a series of cascades,” Desalination, vol. 323, pp. 48–54, 2013.
66
[8]
G. Zaragoza, A. Ruiz-Aguirre, E. Guillen-Burrieza, D. Alarcon-Padilla, and J. Blanco-Galvez,
“Experimental comparison of different prototypes of solar energy driven membrane distillation
systems,” in Proceedings of the 2013 IDA World Congress on Desalination and Water Reuse,
Tianjin, China, October 2013.
[9]
E. Summers and J. H. Lienhard V, “Cycle performance of multi-stage vacuum membrane
distillation (MS-VMD) systems,” in Proceedings of the 2013 IDA World Congress on Desalination
and Water Reuse, Tianjin, China, 2013, Oct. 2013.
[10]
M. Khayet, “Solar desalination by membrane distillation: Dispersion in energy
consumption analysis and water production costs (a review),” Desalination, vol. 308, pp. 89–
101, 2013.
[11]
R. B. Saffarini, E. K. Summers, H. A. Arafat, and J. H. Lienhard V, “Economic evaluation of
stand-alone solar powered membrane distillation systems,” Desalination, vol. 299, pp. 55–62,
2012.
[12]
E. K. Summers and J. H. Lienhard V, “A novel solar-driven air gap membrane distillation
system,” Desalination and Water Treatment, vol. 51, pp. 1–8, 2012.
[13]
A. S. Hassan and H. E. Fath, “Review and assessment of the newly developed MD for
desalination processes,” Desalination and Water Treatment, vol. 51, no. 1-3, pp. 574–585,
2013.
[14]
S. Shirazi, C.-J. Lin, and D. Chen, “Inorganic fouling of pressure-driven membrane
processes – A critical review,” Desalination, vol. 250, no. 1, pp. 236–248, 2010.
67
[15]
H. Susanto, “Towards practical implementations of membrane distillation,” Chemical
Engineering and Processing: Process Intensification, vol. 50, no. 2, pp. 139–150, 2011.
[16]
A. Alklaibi and N. Lior, “Membrane-distillation desalination: Status and potential,”
Desalination, vol. 171, no. 2, pp. 111–131, 2005.
[17]
M. Khayet, “Membranes and theoretical modeling of membrane distillation: A review,”
Advances in Colloid and Interface Science, vol. 164, no. 1-2, pp. 56–88, 2011.
[18]
A. Belmiloudi, Heat Transfer - Theoretical Analysis, Experimental Investigations and
Industrial Systems. InTech, Vienna, 2011.
[19]
A. Antony, J. H. Low, S. Gray, A. E. Childress, P. Le-Clech, and G. Leslie, “Scale formation
and control in high pressure membrane water treatment systems: A review,” Journal of
Membrane Science, vol. 383, no. 1-2, pp. 1–16, 2011.
[20]
R. B. Saffarini, E. K. Summers, H. A. Arafat, and J. H. Lienhard V, “Technical evaluation of
stand-alone solar powered membrane distillation systems,” Desalination, vol. 286, pp. 332–
341, 2012.
[21]
J. Koschikowski, M. Wieghaus, and M. Rommel, “Solar thermal-driven desalination
plants based on membrane distillation,” Desalination, vol. 156, no. 1-3, pp. 295–304, 2003.
[22]
F. Banat, N. Jwaied, M. Rommel, J. Koschikowski, and M. Wieghaus, “Desalination by a
"compact SMADES" autonomous solar powered membrane distillation unit,” Desalination,
vol. 217, no. 1-3, pp. 29–37, 2007.
68
[23]
F. Banat, N. Jwaied, M. Rommel, J. Koschikowski, and M. Wieghaus, “Performance
evaluation of the "large SMADES" autonomous desalination solar-driven membrane distillation
plant in Aqaba, Jordan,” Desalination, vol. 217, no. 1-3, pp. 17–28, 2007.
[24]
E. Guillen-Burrieza, R. Thomas, B. Mansoor, D. Johnson, N. Hilal, and H. Arafat, “Effect of
dry-out on the fouling of PVDF and PTFE membranes under conditions simulating intermittent
seawater membrane distillation (SWMD),” Journal of Membrane Science, vol. 438, pp. 126–139,
2013.
[25]
E. Curcio, X. Ji, G. Di Profio, A. O. Sulaiman, E. Fontananova, and E. Drioli, “Membrane
distillation operated at high seawater concentration factors: Role of the membrane on CaCO3
scaling in presence of humic acid,” Journal of Membrane Science, vol. 346, no. 2, pp. 263–269,
2010.
[26]
M. Gryta, “Effect of iron oxides scaling on the MD process performance,” Desalination,
vol. 216, no. 1-3, pp. 88–102, 2007.
[27]
M. Krivorot, A. Kushmaro, Y. Oren, and J. Gilron, “Factors affecting biofilm formation
and biofouling in membrane distillation of seawater,” Journal of Membrane Science, vol. 376,
no. 1-2, pp. 15–24, 2011.
[28]
M. Gryta, “Polyphosphates used for membrane scaling inhibition during water
desalination by membrane distillation,” Desalination, vol. 285, pp. 170–176, 2012.
[29]
S. Srisurichan, R. Jiraratananon, and A. Fane, “Humic acid fouling in the membrane
distillation process,” Desalination, vol. 174, no. 1, pp. 63–72, 2005.
69
[30]
M. Gryta and M. Tomaszewska, “Membrane distillation of NaCl solution containing
natural organic matter,” Journal of Membrane Science, vol. 181, pp. 279–287, 2001.
[31]
B. Jiao, A. Cassano, and E. Drioli, “Recent advances on membrane processes for the
concentration of fruit juices: A review,” Journal of Food Engineering, vol. 63, pp. 303–324, 2004.
[32]
C. L. Liu, Polygeneration Of Electricity, Heat and Ultrapure Water for the Semiconductor
Industry. PhD thesis, Royal Institute of Technology, Stockholm, Sweden, 2004.
[33]
E. K. Summers, Development of Energy Efficient Membrane Distillation Systems. PhD
thesis, Massachusetts Institute of Technology, Cambridge, Massachusetts, U.S.A., 2013.
[34]
A. S. Al-Amoudi, “Factors affecting natural organic matter (NOM) and scaling fouling in
NF membranes: A review,” Desalination, vol. 259, no. 1-3, pp. 1 – 10, 2010.
[35]
M. Gryta, “Effectiveness of Water Desalination by Membrane Distillation Process,”
Membranes, vol. 2, no. 3, pp. 415–429, 2012.
[36]
F. Morel and J. G. Hering, Principles and Applications of Aquatic Chemistry. Wiley-
Interscience, New York, 1993.
[37]
S. Salvador Cob, C. Beaupin, B. Hofs, M. Nederlof, D. Harmsen, E. Cornelissen,
A. Zwijnenburg, F. Genceli Güner, and G. Witkamp, “Silica and silicate precipitation as limiting
factors in high-recovery reverse osmosis operations,” Journal of Membrane Science, vol. 423424, pp. 1–10, 2012.
70
[38]
T. Waly, M. D. Kennedy, G. J. Witkamp, G. Amy, and J. C. Schippers, “Will calcium
carbonate really scale in seawater reverse osmosis?,” Desalination and Water Treatment, vol. 5,
no. 1-3, pp. 146–152, 2009.
[39]
M. Gryta, “Alkaline scaling in the membrane distillation process,” Desalination, vol. 228,
no. 1-3, pp. 128–134, 2008.
[40]
R. Sheikholeslami, “Scaling potential index (SPI) for CaCO3 based on Gibbs free
energies,” AIChE Journal, vol. 51, no. 6, pp. 1782–1789, 2005.
[41]
K. Al-Anezi and N. Hilal, “Scale formation in desalination plants: Effect of carbon dioxide
solubility,” Desalination, vol. 204, no. 1-3, pp. 385–402, 2007.
[42]
A. M. Shams El Din and R. A. Mohammed, “Brine and scale chemistry in MSF distillers,”
Desalination, vol. 99, pp. 73–111, 1994.
[43]
K. Al-Anezi and N. Hilal, “Effect of carbon dioxide in seawater on desalination: A
comprehensive review,” Separation and Purification Reviews, vol. 35, no. 3, pp. 223–247, 2006.
[44]
J. Ralph and I. Chau, “Mindat - the mineral and locality database.” http://-
www.mindat.org/. Accessed: 2013-10-23.
[45]
N. Andritsos, C. Tzotzi, T. Pahiadaki, S. Yiantsios, and A. Karabelas, “A study of CaCO3
scale formation and inhibition in RO and NF membrane processes,” Journal of Membrane
Science, vol. 296, no. 1-2, pp. 171–84, 2007.
71
[46]
J. Peña, B. Buil, A. Garralon, P. Gomez, M. Turrero, A. Escribano, G. Garralon, and
M. Gomez, “The vaterite saturation index can be used as a proxy of the S&DSI in sea water
desalination by reverse osmosis process,” Desalination, vol. 254, no. 1-3, pp. 75–79, 2010.
[47]
A. G. Xyla, J. Mikroyannidis, and P. G. Koutsoukos, “The inhibition of calcium carbonate
precipitation in aqueous media by organophosphorus compounds,” Journal of Colloid and
Interface Science, vol. 153, no. 2, pp. 537–551, 1992.
[48]
H. Elfil and H. Roques, “Role of hydrate phases of calcium carbonate on the scaling
phenomenon,” Desalination, vol. 137, no. 1-3, pp. 177 – 186, 2001.
[49]
M. Gryta, “The influence of magnetic water treatment on CaCO3 scale formation in
membrane distillation process,” Separation and Purification Technology, vol. 80, no. 2, pp. 293–
299, 2011.
[50]
F. He, K. K. Sirkar, and J. Gilron, “Studies on scaling of membranes in desalination by
direct contact membrane distillation: CaCO3 and mixed CaCO3/CaSO4 systems,” Chemical
Engineering Science, vol. 64, no. 8, pp. 1844–1859, 2009.
[51]
T. Abraham and A. Luthra, “Socio-economic technical assessment of photovoltaic
powered membrane desalination processes for India,” Desalination, vol. 268, no. 1-3, pp. 238 –
248, 2011.
[52]
R. Beck and J.-P. Andreassen, “The onset of spherulitic growth in crystallization of
calcium carbonate,” Journal of Crystal Growth, vol. 312, no. 15, pp. 2226 – 2238, 2010.
72
[53]
M. G. Lioliou, C. A. Paraskeva, P. G. Koutsoukos, and A. C. Payatakes, “Heterogeneous
nucleation and growth of calcium carbonate on calcite and quartz,” Journal of Colloid and
Interface Science, vol. 308, no. 2, pp. 421–428, 2007.
[54]
L.-F. Olsson, “Induction time of precipitation of calcium carbonate,” Journal of Molecular
Liquids, vol. 65, pp. 349–352, 1995.
[55]
O. Pokrovsky, “Precipitation of calcium and magnesium carbonates from homogeneous
supersaturated solutions,” Journal of Crystal Growth, vol. 186, no. 1-2, pp. 233 – 239, 1998.
[56]
X. Yang, R. Tian, S. Ma, and H. Lv, “Study on membrane fouling experiment of stacked
AGMD module in low temperature,” Advanced Materials Research, vol. 396-398, pp. 458–462,
2012.
[57]
M. Gryta, “Long-term performance of membrane distillation process,” Journal of
Membrane Science, vol. 265, no. 1-2, pp. 153–159, 2005.
[58]
M. Gryta, “Calcium sulphate scaling in membrane distillation process,” Chemical Papers,
vol. 63, no. 2, pp. 146–151, 2008.
[59]
K. Karakulski and M. Gryta, “Water demineralisation by NF/MD integrated processes,”
Desalination, vol. 177, no. 1-3, pp. 109 – 119, 2005.
[60]
L. D. Nghiem and T. Cath, “A scaling mitigation approach during direct contact
membrane distillation,” Separation and Purification Technology, vol. 80, pp. 315–322, 2011.
73
[61]
R. Sheikholeslami and M. Ng, “Calcium Sulfate Precipitation in the Presence of
Nondominant Calcium Carbonate: Thermodynamics and Kinetics,” Industrial & Engineering
Chemistry Research, vol. 40, no. 16, pp. 3570–3578, 2001.
[62]
G. Falini, S. Fermani, G. Tosi, and E. Dinelli, “Calcium carbonate morphology and
structure in the presence of seawater ions and humic acids,” Crystal Growth and Design, vol. 9,
no. 5, pp. 2065–2072, 2009.
[63]
A. Hoch, M. Reddy, and G. Aiken, “Calcite crystal growth inhibition by humic substances
with emphasis on hydrophobic acids from the Florida everglades,” Geochimica et Cosmochimica
Acta, vol. 64, no. 1, pp. 61–72, 2000.
[64]
R. Sheikholeslami, “Mixed salts–scaling limits and propensity,” Desalination, vol. 154,
no. 2, pp. 117–127, 2003.
[65]
S. Lee and C.-H. Lee, “Effect of operating conditions on CaSO4 scale formation
mechanism in nanofiltration for water softening,” Water Research, vol. 34, no. 15, pp. 3854–
3866, 2000.
[66]
G. Greenberg, D. Hasson, and R. Semiat, “Limits of RO recovery imposed by calcium
phosphate precipitation,” Desalination, vol. 183, no. 1, pp. 273–288, 2005.
[67]
S. P. Chesters, “Innovations in the inhibition and cleaning of reverse osmosis membrane
scaling and fouling,” Desalination, vol. 238, no. 1, pp. 22–29, 2009.
[68]
A. Zach-Maor, R. Semiat, A. Rahardianto, Y. Cohen, S. Wilson, and S. Gray, “Diagnostic
analysis of RO desalting treated wastewater,” Desalination, vol. 230, no. 1, pp. 239–247, 2008.
74
[69]
R. Ketrane, R. Saidant, O. Gil, L. Leleyter, and F. Baraud, “Efficiency of five scale
inhibitors on calcium carbonate precipitation from hard water: effect of temperature and
concentration,” Desalination, vol. 249, pp. 1397–1404, 2009.
[70]
N. Prihasto, Q. Liu, and S. Kim, “Pre-treatment strategies for seawater desalination by
reverse osmosis system,” Desalination, vol. 249, pp. 308–316, 2009.
[71]
T.-W. Cheng, C.-J. Han, K.-J. Hwang, C.-D. Ho, and W. J. Cooper, “Influence of feed
composition on distillate flux and membrane fouling in direct contact membrane distillation,”
Separation Science and Technology, vol. 45, no. 7, pp. 967–974, 2010.
[72]
M. Gryta, “Desalination of thermally softened water by membrane distillation process,”
Desalination, vol. 257, no. 1-3, pp. 30–35, 2010.
[73]
C. M. Tun, A. G. Fane, J. T. Matheickal, and R. Sheikholeslami, “Membrane distillation
crystallization of concentrated salts-flux and crystal formation,” Journal of Membrane Science,
vol. 257, no. 1-2, pp. 144–155, 2005.
[74]
M. Gryta, “Fouling in direct contact membrane distillation process,” Journal of
Membrane Science, vol. 325, no. 1, pp. 383–394, 2008.
[75]
G. Lange, “Vacuum-driven membrane distillation breakthrough promised by Memsys,”
Desalination and Water Reuse, vol. 20, pp. 23–25, 2011.
[76]
D. Winter, J. Koschikowski, and M. Wieghaus, “Desalination using membrane distillation:
Experimental studies on full scale spiral wound modules,” Journal of Membrane Science,
vol. 375, no. 1-2, pp. 104–112, 2011.
75
[77]
E. Guillén-Burrieza, J. Blanco, G. Zaragoza, D.-C. Alarcón, P. Palenzuela, M. Ibarra, and
W. Gernjak, “Experimental analysis of an air gap membrane distillation solar desalination pilot
system,” Journal of Membrane Science, vol. 379, no. 1-2, pp. 386–396, 2011.
[78]
H.-C. Flemming, G. Schaule, R. McDonogh, and H. F. Ridgway, “Effects and extent of
biofilm accumulation in membrane systems,” Biofouling and Biocorrosion in Industrial Water
Systems, pp. 63–89, 1994.
[79]
Z. Li, V. Yangali-Quintanilla, R. Valladares-Linares, Q. Li, T. Zhan, and G. Amy, “Flux
patterns and membrane fouling propensity during desalination of seawater by forward
osmosis,” WATER RESEARCH, vol. 461, pp. 195–204, 2012.
[80]
L. L. Shreir, Corrosion of Metals and Alloys. John Wiley & Sons, 1963.
[81]
K. Schneider, W. Hölz, R. Wollbeck, and S. Ripperger, “Membranes and modules for
transmembrane distillation,” Journal of membrane science, vol. 39, no. 1, pp. 25–42, 1988.
[82]
A. El-Abbassi, A. Hafidi, M. Khayet, and M. García-Payo, “Integrated direct contact
membrane distillation for olive mill wastewater treatment,” Desalination, vol. 323, pp. 31–38,
2013.
[83]
B. L. Pangarkar, M. G. Sane, and M. Guddad, “Reverse Osmosis and Membrane
Distillation for Desalination of Groundwater: A Review,” ISRN Materials Science, vol. 2011,
pp. 1–9, 2011.
[84]
M. Gryta, “The assessment of microorganism growth in the membrane distillation
system,” Desalination, vol. 142, no. 1, pp. 79–88, 2002.
76
[85]
M. Vogt, H. Flemming, and W. Veeman, “Diffusion in pseudomonas aeruginosa biofilms:
a pulsed field gradient NMR study,” Journal of Biotechnology, vol. 77, pp. 137–146, 2000.
[86]
D. Green and R. Perry, Perry’s Chemical Engineers’ Handbook 8th Edition. McGraw-Hill,
2008.
[87]
G. Meindersma, C. Guijt, and A. de Haan, “Desalination and water recycling by air gap
membrane distillation,” Desalination, vol. 187, no. 1-3, pp. 291–301, 2006.
[88]
H. Zhang, R. Lamb, and J. Lewis, “Engineering nanoscale roughness on hydrophobic
surface–preliminary assessment of fouling behaviour,” Science and Technology of Advanced
Materials, vol. 6, no. 3-4, pp. 236–239, 2005.
[89]
J. Cho, G. Amy, J. Pellegrino, and Y. Yoon, “Characterization of clean and natural organic
matter (NOM) fouled NF and UF membranes, and foulants characterization,” Desalination,
vol. 118, pp. 101–108, 1998.
[90]
J.-M. Laine, C. Campos, I. Baudin, and M.-L. Janex, “Understanding membrane fouling: A
review of over a decade of research,” Water Science and Technology: Water Supply, vol. 3,
no. 5-6, pp. 155 – 164, 2003.
[91]
C. Jucker and M. Clark, “Adsorption of aquatic humic substances on hydrophobic
ultrafiltration membranes,” Journal of Membrane Science, vol. 97, pp. 37–52, 1994.
[92]
M. Yamauchi, S. Katayama, T. Todoroki, and T. Toshio Watanable, “Total synthesis of
fulvic acid,” Journal of the Chemical Society, vol. 5, pp. 1565–1566, 1984.
77
[93]
R. Ishiwatari, “Molecular weight distribution of humic acids from lake and marine
sediments,” Geochemical Journal, vol. 5, pp. 121–132, 1971.
[94]
M. Khayet, J. Velazque, and Mengual, “Direct contact membrane distillation of humic
acid solutions,” Journal of Membrane Science, vol. 240, pp. 123–128, 2004.
[95]
S. Meng, Y. Yun, J. Mansouri, and V. Chen, “Fouling and crystallisation behaviour of
superhydrophobic nano-composite PVDF membranes in direct contact membrane distillation,”
Journal of Membrane Science, vol. 463, pp. 102–112, 2014.
[96]
W. Yuan and A. Zydney, “Humic acid fouling during microfiltration,” Journal of
Membrane Science, vol. 157, pp. 1–12, 1999.
[97]
K. Zhao, W. Heinzl, M. Wenzel, S. Büttner, F. Bollen, G. Lange, S. Heinzl, and N. Sarda,
“Experimental study of the memsys vacuum-multi-effect-membrane-distillation (V-MEMD)
module,” Desalination, vol. 323, pp. 150–160, 2013.
[98]
S. Hong and M. Elimelech, “Chemical and physical aspects of natural organic matter
(NOM) fouling of nanofiltration membranes,” Journal of membrane science, vol. 132, no. 2,
pp. 159–181, 1997.
[99]
S. Minegishi, K. Iketa, Y. Watanabe, and H. Yamamura, “Main foulants analyses of
hollow fiber membrane during UF filtration,” Japanese Water Works Association, vol. 71,
p. 812, 2002.
78
[100] Y. Hao, A. Moriya, T. Maruyama, Y. Ohmukai, and H. Matsuyama, “Effect of metal ions
on humic acid fouling of hollow fiber ultrafiltration membrane,” Journal of Membrane Science,
vol. 376, pp. 247–253, 2011.
[101] L. D. Nghiem and T. Cath, “A scaling mitigation approach during direct contact
membrane distillation,” Separation and Purification Technology, vol. 80, no. 2, pp. 315–322,
2011.
[102] M. Gryta, M. Tomaszewska, and K. Karakulski, “Wastewater treatment by membrane
distillation,” Desalination, vol. 198, no. 1-3, pp. 67–73, 2006.
[103] J. Ortiz de Zarate, C. Rincon, and J. Mengual, “Concentration of bovine serum albumin
aqueous solutions by membrane distillation,” Separation Science and Technology, vol. 33,
pp. 283–296, 1998.
[104] J. Xu, S. Lange, J. Bartley, and R. Johnson, “Alginate-coated microporous PTFE
membranes for use in the osmotic distillation of oily feeds,” Journal of Membrane Science,
vol. 240, pp. 81–89, 2004.
[105] M. Gryta, “Fouling in direct contact membrane distillation process,” Journal of
Membrane Science, vol. 325, no. 1, pp. 383–394, 2008.
[106] R. W. Schofield, A. Fane, and C. Fell, “Gas and vapour transport through microporous
membranes in Knudsen-Poiseuille transition,” Journal of Membrane Science, vol. 53, pp. 159–
171, 1990.
79
[107] K. H. Mistry, A. Mitsos, and J. H. Lienhard, “Optimal operating conditions and
configurations for humidification–dehumidification desalination cycles,” International Journal
of Thermal Sciences, vol. 50, no. 5, pp. 779–789, 2011.
[108] L. Dudley and E. Darton, “Pretreatment procedures to control biogrowth and scale
formation in membrane systems,” Desalination, vol. 110, no. 1-2, pp. 11–20, 1997.
[109] J.-P. Mericq, S. Laborie, and C. Cabassud, “Vacuum membrane distillation of seawater
reverse osmosis brines,” Water Research, vol. 44, no. 18, pp. 5260–73, 2010.
[110] M. Gryta, “Water Desalination by Membrane Distillation,” Desalination, Trends and
Technologies, pp. 21–40, 2011.
[111] M. Gryta, “Influence of polypropylene membrane surface porosity on the performance
of membrane distillation process,” Journal of Membrane Science, vol. 287, no. 1, pp. 67–78,
2007.
[112] M. Gryta, J. Grzechulska-Damszel, a. Markowska, and K. Karakulski, “The influence of
polypropylene degradation on the membrane wettability during membrane distillation,”
Journal of Membrane Science, vol. 326, no. 2, pp. 493–502, 2009.
[113] S. Hsu, K. Cheng, and J. Chiou, “Seawater desalination by direct contact membrane
distillation,” Desalination, vol. 143, no. 3, pp. 279–287, 2002.
[114] V. Gekas and B. Hallstrom, “Mass Transfer in the Membrane Concentration Polarization
Layer under Turbulent Cross Flow,” Journal of Membrane Science, vol. 30, pp. 153–170, 1987.
80
[115] F. Lagana, G. Barbieri, and E. Drioli, “Direct contact membrane distillation: Modelling
and concentration experiments,” Journal of Membrane Science, vol. 166, no. 1, pp. 1 – 11, 2000.
[116] E. K. Summers, H. A. Arafat, and J. H. Lienhard, “Energy efficiency comparison of singlestage membrane distillation (MD) desalination cycles in different configurations,” Desalination,
vol. 290, pp. 54–66, 2012.
[117] T. Van Gassel and K. Schneider, “An energy-efficient membrane distillation process,”
Membranes and Membrane Processes, pp. 343–348, 1986.
[118] K. He, H. J. Hwang, M. W. Woo, and I. S. Moon, “Production of drinking water from
saline water by direct contact membrane distillation (DCMD),” Journal of Industrial and
Engineering Chemistry, vol. 17, no. 1, pp. 41–48, 2011.
[119] S. Srisurichan, R. Jiraratananon, and A. Fane, “Mass transfer mechanisms and transport
resistances in direct contact membrane distillation process,” Journal of membrane science,
vol. 277, pp. 186–194, 2006.
[120] A. Rohatgi, “Web plot digitizer.” http://arohatgi.info/WebPlotDigitizer/, 2010-2013.
[121] B. L. Pangarkar, M. G. Sane, S. B. Parjane, and M. Guddad, “Vacuum Membrane
Distillation for Desalination of Ground Water by using Flat Sheet Membrane,” International
Journal of Chemical and Biological Engineering, pp. 13–18, 2011.
[122] S. Agashichev and A. Sivakov, “Modeling and calculation of temperature-concentration
polarisation in the membrane distillation process (MD),” Desalination, vol. 93, no. 1-3, pp. 245–
258, 1993.
81
[123] S. Goh, J. Zhang, Y. Liu, and A. Fane, “Fouling and wetting in membrane distillation (MD)
and MD-bioreactor (MDBR) for wastewater reclamation,” Desalination, vol. 323, pp. 39–47,
2013.
[124] A. Hausmann, P. Sanciolo, T. Vasiljevic, U. Kulozik, and M. Duke, “Performance
assessment of membrane distillation for skim milk and whey processing,” Journal Dairy of
Science, vol. 97, pp. 56–71, 2014.
[125] A. Razmjou,
E. Arifin,
G. Dong,
J. Mansouri,
and
V. Chen,
“Superhydrophobic
modification of TiO2 nanocomposite PVDF membranes for applications in membrane
distillation,” Journal of Membrane Science, vol. 415-416, pp. 850–863, 2012.
[126] K. Karakulski, M. Gryta, and A. Morawski, “Membrane processes used for potable water
quality improvement,” Desalination, vol. 145, no. 1-3, pp. 315–319, 2002.
[127] U. Kesieme, N. Milne, H. Aral, C. Cheng, and M. Duke, “Economic analysis of desalination
technologies in the context of carbon pricing, and opportunities for membrane distillation,”
Desalination, vol. 323, pp. 66–74, 2013.
[128] E. Drioli, Y. Wu, and V. Calabro, “Membrane distillation in the treatment of aqueous
solutions,” Journal of Membrane Science, vol. 33, no. 3, pp. 277–284, 1987.
[129] V. Calabro, B. L. Jiao, and E. Drioli, “Theoretical and experimental study on membrane
distillation in the concentration of orange juice,” Industrial & Engineering Chemistry Research,
vol. 33, no. 7, pp. 1803–1808, 1994.
82
[130] A. Alklaibi and N. Lior, “Heat and mass transfer resistance analysis of membrane
distillation,” Journal of Membrane Science, vol. 282, no. 1-2, pp. 362–369, 2006.
[131] K. Al-Anezi, C. Somerfield, D. Mee, N. Hankins, and N. Hilal, “Effect of anti-scale agents
on the solubility of CO2 in seawater at temperatures of 60to 90and pressures of 1–2 bar,”
Desalination, vol. 227, no. 1-3, pp. 46–56, 2008.
[132] M. Malki, “Optimizing scale inhibition costs in reverse osmosis desalination plant,”
International Desalination And Water Reuse Quarterly, vol. 17, no. 4, p. 28, 2008.
[133] G. Rabin, “Scale control in saline and wastewater desalination,” Israel Journal of
Chemistry, vol. 46, pp. 97–104, 2006.
[134] A. Franken, J. Nolten, M. Mulder, D. Bargeman, and C. Smolders, “Wetting criteria for
the applicability of membrane distillation,” Journal of Membrane Science, vol. 33, pp. 315–328,
1987.
[135] F. He, K. K. Sirkar, and J. Gilron, “Effects of antiscalants to mitigate membrane scaling by
direct contact membrane distillation,” Journal of Membrane Science, vol. 345, no. 1-2, pp. 53–
58, 2009.
[136] J. van der Hoek, J. Hofman, P. Bonné, M. Nederlov, and H. Vrouwenvelder, “RO
treatment: selection of a pretreatment scheme based on fouling characteristics and operating
conditions based on environmental impact,” Desalination, vol. 127, pp. 89–101, 1999.
[137] N. Yang, “Physical Conditioning for Scale Prevention during Desalination by Reverse
Osmosis,” Master’s Thesis: Chalmers Institute of Technology, Goteborg, Sweden, 2005.
83
[138] E. Drioli, E. Curcio, A. Criscuoli, and G. D. Profio, “Integrated system for recovery of
CaCO3, NaCl and MgSO4*H2O from nanofiltration retentate,” Journal of Membrane Science,
vol. 239, no. 1, pp. 27–38, 2004.
[139] W. Schnabel, Polymer degradation: principles and practical applications. Hanser
International, 1981.
[140] M. Tomaszewska, M. Gryta, and A. Morawski, “Recovery of hydrochloric acid from metal
pickling solutions by membrane distillation,” Separation and Purification Technology, vol. 22,
pp. 591–600, 2001.
[141] L. Song, B. Li, K. K. Sirkar, and J. L. Gilron, “Direct Contact Membrane Distillation-Based
Desalination: Novel Membranes, Devices, Larger-Scale Studies, and a Model,” Industrial &
Engineering Chemistry Research, vol. 46, no. 8, pp. 2307–2323, 2007.
[142] Z. Ma, Y. Hong, L. Ma, and M. Su, “Superhydrophobic membranes with ordered arrays of
nanospiked microchannels for water desalination,” Langmuir : the ACS journal of surfaces and
colloids, vol. 25, no. 10, pp. 5446–50, 2009.
[143] F. He, J. Gilron, H. Lee, L. Song, and K. K. Sirkar, “Potential for scaling by sparingly soluble
salts in crossflow DCMD,” Journal of Membrane Science, vol. 311, no. 1-2, pp. 68–80, 2008.
[144] Y.-F. Yang, Y. Li, Q.-L. Li, L.-S. Wan, and Z.-K. Xu, “Surface hydrophilization of
microporous polypropylene membrane by grafting zwitterionic polymer for anti-biofouling,”
Journal of Membrane Science, vol. 362, pp. 81–89, 2010.
84
[145] Y.-F. Yang, L.-S. Wan, and Z.-K. Xu, “Surface hydrophilization of microporous
polypropylene membrane by the interfacial crosslinking of polyethylenimine,” Journal of
Membrane Science, vol. 337, pp. 70–80, 2009.
[146] E. Curcio and E. Drioli, “Separation & Purification Reviews Membrane Distillation and
Related Operations – A Review,” Separation and Purification Reviews, vol. 34, pp. 35–86, 2005.
[147] A. Alklaibi and N. Lior, “Comparative study of direct-contact and air-gap membrane
distillation processes,” Industrial & engineering chemistry research, pp. 584–590, 2007.
[148] J. Swaminathan, “Numerical and experimental investigation of membrane distillation
flux and energy efficiency,” Master’s thesis, Massachusetts Institute of Technology, 2014.
[149] G. Thiel, S. Zubair, and J. Lienhard V, “An analysis of likely scalants in the treatment of
produced water from Nova Scotia,” Heat Transfer Engineering, 2013.
[150] G. Chen, X. Yang, R. Wang, and A. G. Fane, “Performance enhancement and scaling
control with gas bubbling in direct contact membrane distillation,” Desalination, vol. 308,
pp. 47–55, 2013.
[151] K. Sakai, T. Koyano, T. Muroi, and M. Tamura, “Effects of temperature and concentration
polarization on water vapour permeability for blood in membrane distillation,” The Chemical
Engineering Journal, vol. 38, no. 3, pp. B33–B39, 1988.
[152] M. Gryta, K. Karakulski, M. Tomaszewska, and A. Morawski, “Treatment of effluents
from the regeneration of ion exchangers using the MD process,” Desalination, vol. 180, no. 1,
pp. 173–180, 2005.
85
[153] F. Banat, S. Al-Asheh, and M. Qtaishat, “Treatment of waters colored with methylene
blue dye by vacuum membrane distillation,” Desalination, vol. 174, no. 1, pp. 87–96, 2005.
86